# Fusion Accounting - OFX Bank Statement Parser
# Original implementation for Open Financial Exchange v1 (SGML) and v2 (XML)
# Based on the published OFX specification (https://www.ofx.net/spec)

import logging
import re
from datetime import datetime
from xml.etree import ElementTree

from odoo import _, models
from odoo.exceptions import UserError

_log = logging.getLogger(__name__)


class FusionOFXParser:
    """Standalone parser for OFX (Open Financial Exchange) files.

    Supports both OFX v1 (SGML-like markup without closing tags) and
    OFX v2 (well-formed XML).  The parser normalises either dialect into
    a common intermediate structure before extracting statement data.

    This is an **original** implementation written from the published
    OFX 1.6 / 2.2 specification — it is not derived from Odoo Enterprise.
    """

    # OFX date format: YYYYMMDDHHMMSS[.XXX[:TZ]]  — timezone and fractional
    # seconds are optional; many banks only emit YYYYMMDD.
    _OFX_DATE_RE = re.compile(
        r'^(\d{4})(\d{2})(\d{2})'       # YYYYMMDD (required)
        r'(?:(\d{2})(\d{2})(\d{2}))?'    # HHMMSS   (optional)
        r'(?:\.\d+)?'                     # .XXX     (optional fractional)
        r'(?:\[.*\])?$'                   # [:TZ]    (optional timezone)
    )

    # SGML self-closing tags used in OFX v1 (no closing tag counterpart).
    # These contain scalar data directly after the tag.
    _SGML_LEAF_TAGS = {
        'TRNTYPE', 'DTPOSTED', 'DTUSER', 'DTSTART', 'DTEND',
        'TRNAMT', 'FITID', 'CHECKNUM', 'REFNUM', 'NAME', 'MEMO',
        'PAYEEID', 'ACCTID', 'BANKID', 'BRANCHID', 'ACCTTYPE',
        'BALAMT', 'DTASOF', 'CURDEF', 'SEVERITY', 'CODE', 'MESSAGE',
        'SIC', 'PAYEEID', 'CORRECTFITID', 'CORRECTACTION',
        'SRVRTID', 'CLRTID',
    }

    # -------------------------------------------------------------------
    # Public API
    # -------------------------------------------------------------------

    def parse_ofx(self, data_file):
        """Parse an OFX file (bytes or str) and return a list of statement
        dicts compatible with the Fusion Accounting import pipeline.

        Each dict has the keys:
            - ``name``            : statement identifier
            - ``date``            : closing date (datetime.date)
            - ``balance_start``   : opening balance (float)
            - ``balance_end_real``: closing balance (float)
            - ``currency_code``   : ISO 4217 currency code
            - ``account_number``  : bank account number
            - ``transactions``    : list of transaction dicts

        Transaction dicts contain:
            - ``date``              : posting date (datetime.date)
            - ``payment_ref``       : description / memo
            - ``ref``               : FITID or reference number
            - ``amount``            : signed float (negative = debit)
            - ``unique_import_id``  : unique per-transaction identifier
            - ``transaction_type``  : OFX TRNTYPE value
        """
        raw = self._to_text(data_file)

        # Determine OFX dialect and obtain an ElementTree root
        if self._is_ofx_v2(raw):
            root = self._parse_xml(raw)
        else:
            root = self._parse_sgml(raw)

        return self._extract_statements(root)

    # -------------------------------------------------------------------
    # Input normalisation
    # -------------------------------------------------------------------

    @staticmethod
    def _to_text(data_file):
        """Ensure *data_file* is a string, decoding bytes if necessary."""
        if isinstance(data_file, bytes):
            # Try UTF-8 first; fall back to Latin-1 (lossless for any byte)
            for encoding in ('utf-8-sig', 'utf-8', 'latin-1'):
                try:
                    return data_file.decode(encoding)
                except UnicodeDecodeError:
                    continue
        return data_file

    @staticmethod
    def _is_ofx_v2(text):
        """Return True when *text* looks like OFX v2 (XML) rather than
        SGML-based v1.  OFX v2 begins with an XML processing instruction
        or a ``<?OFX …?>`` header."""
        stripped = text.lstrip()
        return stripped.startswith('<?xml') or stripped.startswith('<?OFX')

    # -------------------------------------------------------------------
    # OFX v2 (XML) parser
    # -------------------------------------------------------------------

    def _parse_xml(self, text):
        """Parse well-formed OFX v2 XML and return the root Element."""
        try:
            return ElementTree.fromstring(text.encode('utf-8'))
        except ElementTree.ParseError as exc:
            raise UserError(
                _("Failed to parse OFX XML file: %s", str(exc))
            ) from exc

    # -------------------------------------------------------------------
    # OFX v1 (SGML) parser — convert to XML then parse
    # -------------------------------------------------------------------

    def _parse_sgml(self, text):
        """Convert an OFX v1 SGML document into well-formed XML and
        return the root Element.

        The SGML dialect used by OFX v1 has two kinds of tags:
        *   **Aggregate** tags like ``<STMTTRNRS>`` which contain child
            elements and always have a matching ``</STMTTRNRS>``.
        *   **Leaf** (data) tags like ``<TRNAMT>-42.50`` which carry a
            scalar value and are never explicitly closed.

        The conversion strategy inserts explicit close tags for every
        leaf element so that the result is valid XML.
        """
        # Strip the SGML headers (everything before the first ``<OFX>``).
        ofx_idx = text.upper().find('<OFX>')
        if ofx_idx == -1:
            raise UserError(_("The file does not contain a valid OFX document."))
        body = text[ofx_idx:]

        # Normalise whitespace inside tags: collapse runs of whitespace
        # between ``>`` and ``<`` but preserve data values.
        lines = body.splitlines()
        xml_lines = []

        for line in lines:
            stripped = line.strip()
            if not stripped:
                continue
            xml_lines.append(stripped)

        joined = '\n'.join(xml_lines)

        # Insert closing tags for leaf elements.
        # A leaf tag looks like ``<TAGNAME>value`` (no ``</TAGNAME>`` follows).
        def _close_leaf_tags(sgml_text):
            """Insert ``</TAG>`` after each leaf tag's data value."""
            result = []
            tag_re = re.compile(r'<(/?)(\w+)>(.*)', re.DOTALL)
            for raw_line in sgml_text.split('\n'):
                raw_line = raw_line.strip()
                if not raw_line:
                    continue
                m = tag_re.match(raw_line)
                if m:
                    is_close = m.group(1) == '/'
                    tag_name = m.group(2).upper()
                    rest = m.group(3).strip()

                    if is_close:
                        result.append(f'</{tag_name}>')
                    elif tag_name in self._SGML_LEAF_TAGS:
                        # Leaf element: value sits between open and (missing) close tag
                        data_val = rest.split('<')[0].strip() if '<' in rest else rest
                        result.append(f'<{tag_name}>{self._xml_escape(data_val)}</{tag_name}>')
                        # If the rest of the line has another tag, process it
                        if '<' in rest:
                            leftover = rest[rest.index('<'):]
                            for extra in _close_leaf_tags(leftover).split('\n'):
                                if extra.strip():
                                    result.append(extra.strip())
                    else:
                        # Aggregate (container) tag — keep as-is
                        result.append(f'<{tag_name}>')
                        if rest:
                            for extra in _close_leaf_tags(rest).split('\n'):
                                if extra.strip():
                                    result.append(extra.strip())
                else:
                    result.append(raw_line)
            return '\n'.join(result)

        xml_text = _close_leaf_tags(joined)

        try:
            return ElementTree.fromstring(xml_text.encode('utf-8'))
        except ElementTree.ParseError as exc:
            _log.debug("SGML→XML conversion result:\n%s", xml_text[:2000])
            raise UserError(
                _("Failed to parse OFX v1 (SGML) file. The file may be "
                  "corrupt or in an unsupported dialect: %s", str(exc))
            ) from exc

    @staticmethod
    def _xml_escape(text):
        """Escape XML-special characters in *text*."""
        return (
            text.replace('&', '&amp;')
                .replace('<', '&lt;')
                .replace('>', '&gt;')
                .replace('"', '&quot;')
                .replace("'", '&apos;')
        )

    # -------------------------------------------------------------------
    # Data extraction
    # -------------------------------------------------------------------

    def _extract_statements(self, root):
        """Walk the parsed OFX element tree and collect statement data.

        Supports ``BANKMSGSRSV1`` (bank accounts) and ``CCMSGSRSV1``
        (credit-card accounts).
        """
        statements = []

        # Locate all statement response containers
        for tag_suffix, acct_tag in [
            ('BANKMSGSRSV1', 'BANKACCTFROM'),
            ('CCMSGSRSV1', 'CCACCTFROM'),
        ]:
            for stmtrs in self._find_all(root, 'STMTRS') + self._find_all(root, 'CCSTMTRS'):
                stmt = self._extract_single_statement(stmtrs, acct_tag)
                if stmt:
                    statements.append(stmt)

        if not statements:
            raise UserError(
                _("No bank or credit-card statements found in the OFX file.")
            )
        return statements

    def _extract_single_statement(self, stmtrs, acct_tag):
        """Extract one statement from a ``<STMTRS>`` or ``<CCSTMTRS>``
        element."""
        # Currency
        currency = self._find_text(stmtrs, 'CURDEF') or ''

        # Account number
        acct_elem = self._find_first(stmtrs, acct_tag)
        if acct_elem is None:
            acct_elem = self._find_first(stmtrs, 'BANKACCTFROM')
        if acct_elem is None:
            acct_elem = self._find_first(stmtrs, 'CCACCTFROM')

        acct_number = ''
        if acct_elem is not None:
            acct_number = self._find_text(acct_elem, 'ACCTID') or ''

        # Transaction list
        txn_list_el = self._find_first(stmtrs, 'BANKTRANLIST')
        if txn_list_el is None:
            txn_list_el = stmtrs  # CCSTMTRS may put transactions directly inside

        start_date = self._parse_ofx_date(self._find_text(txn_list_el, 'DTSTART'))
        end_date = self._parse_ofx_date(self._find_text(txn_list_el, 'DTEND'))

        transactions = []
        for stmttrn in self._find_all(txn_list_el, 'STMTTRN'):
            txn = self._extract_transaction(stmttrn)
            if txn:
                transactions.append(txn)

        # Balances — look for LEDGERBAL and AVAILBAL
        balance_start = 0.0
        balance_end = 0.0

        ledger_bal = self._find_first(stmtrs, 'LEDGERBAL')
        if ledger_bal is not None:
            balance_end = self._safe_float(self._find_text(ledger_bal, 'BALAMT'))

        avail_bal = self._find_first(stmtrs, 'AVAILBAL')
        if avail_bal is not None and ledger_bal is None:
            balance_end = self._safe_float(self._find_text(avail_bal, 'BALAMT'))

        # Derive opening balance:  opening = closing − sum(transactions)
        txn_total = sum(t['amount'] for t in transactions)
        balance_start = balance_end - txn_total

        stmt_date = end_date or (start_date if start_date else None)
        stmt_name = f"OFX {acct_number}" if acct_number else "OFX Import"
        if stmt_date:
            stmt_name += f" {stmt_date.strftime('%Y-%m-%d')}"

        return {
            'name': stmt_name,
            'date': stmt_date,
            'balance_start': balance_start,
            'balance_end_real': balance_end,
            'currency_code': currency.upper() if currency else None,
            'account_number': acct_number,
            'transactions': transactions,
        }

    def _extract_transaction(self, stmttrn):
        """Extract a single transaction from a ``<STMTTRN>`` element."""
        trntype = self._find_text(stmttrn, 'TRNTYPE') or ''
        dt_posted = self._parse_ofx_date(self._find_text(stmttrn, 'DTPOSTED'))
        dt_user = self._parse_ofx_date(self._find_text(stmttrn, 'DTUSER'))
        amount = self._safe_float(self._find_text(stmttrn, 'TRNAMT'))
        fitid = self._find_text(stmttrn, 'FITID') or ''
        checknum = self._find_text(stmttrn, 'CHECKNUM') or ''
        refnum = self._find_text(stmttrn, 'REFNUM') or ''
        name = self._find_text(stmttrn, 'NAME') or ''
        memo = self._find_text(stmttrn, 'MEMO') or ''

        # Build description: prefer NAME, append MEMO if different
        description = name
        if memo and memo != name:
            description = f"{name} - {memo}" if name else memo

        # Build reference: FITID is the primary unique ID; CHECKNUM or REFNUM
        # serve as human-readable reference
        ref = checknum or refnum or fitid
        unique_id = fitid

        return {
            'date': dt_user or dt_posted,
            'payment_ref': description or ref or '/',
            'ref': ref,
            'amount': amount,
            'unique_import_id': unique_id,
            'transaction_type': trntype,
        }

    # -------------------------------------------------------------------
    # Element-tree helpers (case-insensitive tag search)
    # -------------------------------------------------------------------

    @staticmethod
    def _find_all(parent, tag):
        """Find all descendant elements whose tag matches *tag*
        (case-insensitive)."""
        tag_upper = tag.upper()
        return [el for el in parent.iter() if el.tag.upper() == tag_upper]

    @staticmethod
    def _find_first(parent, tag):
        """Return the first descendant matching *tag* (case-insensitive)
        or ``None``."""
        tag_upper = tag.upper()
        for el in parent.iter():
            if el.tag.upper() == tag_upper:
                return el
        return None

    @classmethod
    def _find_text(cls, parent, tag):
        """Return stripped text content of the first descendant matching
        *tag*, or ``None``."""
        el = cls._find_first(parent, tag)
        if el is not None and el.text:
            return el.text.strip()
        return None

    # -------------------------------------------------------------------
    # Date / numeric helpers
    # -------------------------------------------------------------------

    @classmethod
    def _parse_ofx_date(cls, date_str):
        """Parse an OFX date string (``YYYYMMDD…``) into a Python date."""
        if not date_str:
            return None
        m = cls._OFX_DATE_RE.match(date_str.strip())
        if not m:
            # Fallback: try basic YYYYMMDD
            try:
                return datetime.strptime(date_str.strip()[:8], '%Y%m%d').date()
            except (ValueError, IndexError):
                _log.warning("Unparseable OFX date: %s", date_str)
                return None
        year, month, day = int(m.group(1)), int(m.group(2)), int(m.group(3))
        try:
            return datetime(year, month, day).date()
        except ValueError:
            _log.warning("Invalid OFX date components: %s", date_str)
            return None

    @staticmethod
    def _safe_float(value):
        """Convert *value* to float, returning 0.0 for empty / invalid."""
        if not value:
            return 0.0
        try:
            return float(value.replace(',', '.'))
        except (ValueError, AttributeError):
            return 0.0


class FusionJournalOFXImport(models.Model):
    """Register OFX as an available bank-statement import format and
    implement the parser hook on ``account.journal``."""

    _inherit = 'account.journal'

    # ---- Format Registration ----
    def _get_bank_statements_available_import_formats(self):
        """Append OFX to the list of importable formats."""
        formats = super()._get_bank_statements_available_import_formats()
        formats.append('OFX')
        return formats

    # ---- Parser Hook ----
    def _parse_bank_statement_file(self, attachment):
        """Attempt to parse *attachment* as OFX.  Falls through to
        ``super()`` when the file is not recognised as OFX."""
        raw_data = attachment.raw
        if not self._is_ofx_file(raw_data):
            return super()._parse_bank_statement_file(attachment)

        parser = FusionOFXParser()
        try:
            statements = parser.parse_ofx(raw_data)
        except UserError:
            raise
        except Exception as exc:
            _log.exception("OFX parsing error")
            raise UserError(
                _("Could not parse the OFX file: %s", str(exc))
            ) from exc

        # The import pipeline expects (currency_code, account_number, stmts)
        currency_code = None
        account_number = None
        if statements:
            currency_code = statements[0].get('currency_code')
            account_number = statements[0].get('account_number')

        return currency_code, account_number, statements

    # ---- Detection ----
    @staticmethod
    def _is_ofx_file(raw_data):
        """Heuristic check: does *raw_data* look like an OFX file?"""
        try:
            text = raw_data.decode('utf-8-sig', errors='ignore')[:4096]
        except (UnicodeDecodeError, AttributeError):
            text = str(raw_data)[:4096]
        text_upper = text.upper()
        # OFX v2 (XML)
        if '<?OFX' in text_upper or '<OFX>' in text_upper:
            return True
        # OFX v1 (SGML header markers)
        if 'OFXHEADER:' in text_upper:
            return True
        return False