from __future__ import annotations

import io
import re
import unicodedata as ud
from pathlib import Path
from typing import Dict, Union, Any

import pandas as pd
import pdfplumber

###############################################################################
# Normalização                                                                #
###############################################################################
_DASH_CHARS = "\u2010\u2011\u2012\u2013\u2014\u2212"
_NBSP_CHARS = "\u00A0\u202F\u2007"
_dash_trans = str.maketrans({c: "-" for c in _DASH_CHARS})
_nbsp_trans = str.maketrans({c: " " for c in _NBSP_CHARS})


def _normalize_line(s: str) -> str:
    s = s.translate(_dash_trans).translate(_nbsp_trans)
    s = ud.normalize("NFKC", s)
    return re.sub(r"\s{2,}", " ", s.strip())


def _parse_money(value: str) -> float | None:
    value = value.strip()
    if value == "-":
        return None
    return float(value.replace(".", "").replace(",", "."))

###############################################################################
# Regex – Cabeçalhos                                                          #
###############################################################################
RX_MSISDN_HEADER = re.compile(r"Vantagens que seu plano oferece:\s*(?P<msisdn>\d{2}\s\d{5}-\d{4})", re.I)
RX_MSISDN_SEU_NUM = re.compile(r"SEU\s+NÚMERO\s+TIM\s+(?P<msisdn>\d{2}\s\d{5}-\d{4})", re.I)
RX_MSISDN_DETALHE = re.compile(r"Detalhamento de Serviços\s+N[°º]\s*(?P<msisdn>\d{2}\s\d{5}-\d{4})", re.I)

SECTION_HEADERS: Dict[str, str] = {
    "IGNORE Ilimitados": r"^Detalhamento de Serviços Ilimitados",
    "IGNORE Detalhamento": r"^Detalhamento de Serviç[io]s\b",
    "Fatura Resumo": r"^FATURA\s+RESUMO",
    "DANFE-COM": r"^DANFE-COM\b",
    "Plano": r"^Plano\b",
    "Mensalidades Adicionais": r"^MENSALIDADES\s+ADICIONAIS",
    "Itens Eventuais": r"^ITENS\s+EVENTUAIS",
    "TIM Viagem": r"^TIM\s+VIAGEM",
    "SVA Detalhe Total": r"^Serviços\s+de\s+Valor\s+Adicionado\s+Total",
    "Desconto Franquia": r"^Desconto\(s\)\s+Franquia",
    "Desconto SVA": r"^Desconto\(s\)\s+Serviç[io]s",
    "Franquia": r"^Franquia\s*\(s\)",
    "SVA": r"^Serviços\s+de\s+valor\s+adicionado\(SVA\)",
    "Chamadas Rede TIM": r"^CHAMADAS\s+DENTRO\s+DA\s+REDE\s+TIM",
    "Chamadas Fora Rede TIM": r"^CHAMADAS\s+FORA\s+DA\s+REDE\s+TIM",
    "Outros Valores": r"^OUTROS\s+VALORES",
    "Deduções": r"^DEDUÇÕES",
    "Roaming Internacional": r"^ROAMING\s+INTERNACIONAL",
    "Cobranças de Terceiros": r"^COBRANÇAS\s+DE\s+TERCEIROS",
    "Débitos de outras operadoras": r"^DÉBITOS\s+DE\s+OUTRAS\s+OPERADORAS"
}
SECTION_REGEX = {k: re.compile(v, re.I) for k, v in SECTION_HEADERS.items()}
RX_SECTION_TOTAL = re.compile(r"(?:R\$\s*)?(-?[\d\.]+,\d{2})")

# Patterns que encerram a seção atual sem iniciar uma nova
RX_SECTION_BREAK = re.compile(
    r"^(Nota Fiscal de Servi|SEUS\s+DADOS|ITEM\s+QTDE\s+ICMS|TOTAL\s+TIM|"
    r"Ficou\s+com\s+dúvidas|Bancos\s+Conveniados|Reservado\s+ao\s+fisco|"
    r"Tipo:\s+N\s+-\s+Normal)",
    re.I,
)

###############################################################################
# Regex – Linhas                                                              #
###############################################################################
RX_PERIOD = r"(?P<period>(?:\d{2}/\d{2}\s+a\s+\d{2}/\d{2}|-))"
RX_DAYS = r"(?P<days>(?:\d+|-))"
RX_VALUE = r"(?P<value>-?\d+,\d{2}|Incluído)"
RX_PARCEL = r"(?P<parcel>(?:\d+/\d+|-))"

# --- Plano / Desconto / Chamadas (layout padrão: QTY DESC PARCELA PERIOD DIAS VALOR) ---
RX_SIMPLE = re.compile(rf"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)\s+{RX_PARCEL}\s+{RX_PERIOD}\s+{RX_DAYS}\s+{RX_VALUE}\s*$")
RX_CONSUMPTION = re.compile(rf"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)\s+{RX_PARCEL}\s+(?P<franchise>Ilimitado|-)\s+(?P<consumption>\d{{1,3}}m\d{{2}}s)\s+{RX_PERIOD}\s+{RX_DAYS}\s+{RX_VALUE}\s*$")
RX_SUBTOTAL = re.compile(r"^Subtotal\s+(?P<value>-?\d+,\d{2})\s*$", re.I)
RX_DETAIL   = re.compile(r"^\d+\s+.+?\s+(-?\d+,\d{2})\s*$")
# Fatura Resumo
RX_RESUMO = re.compile(r"^\s*(?P<desc>.+?)\s+R\$\s*(?P<value>-?\d+,\d{2})\s*$")
RX_TOTAL_GERAL = re.compile(r"^Total\s+geral\s+R\$\s*(?P<value>-?\d+,\d{2})", re.I)
RX_FATURA_METADATA = re.compile(
    r"FATURA\s+PER[ÍI]ODO\s+EMISS[ÃA]O\s+POSTAGEM\s+"
    r"(?P<fatura>\S+)\s+"
    r"(?P<period>\d{2}/\d{2}\s+a\s+\d{2}/\d{2})\s+"
    r"(?P<emissao>\d{2}/\d{2}/\d{4})\s+"
    r"(?P<postagem>\d{2}/\d{2}/\d{4})",
    re.I,
)
RX_DANFE_TOTAL = re.compile(r"^Total\s+geral\s+R\$\s*(?P<value>-?\d+(?:\.\d{3})*,\d{2})", re.I)
RX_DANFE_ROW = re.compile(
    r"^(?P<desc>.+?)\s+"
    r"(?P<unit>[A-Z]{2})\s+"
    r"(?P<qty>\d+(?:,\d+)?)\s+"
    r"(?P<preco_unit>-?\d+(?:\.\d{3})*,\d{2})\s+"
    r"(?P<pis_cofins>-?\d+(?:\.\d{3})*,\d{2}|-)\s+"
    r"(?P<bc_icms>-?\d+(?:\.\d{3})*,\d{2}|-)\s+"
    r"(?P<aliq_icms>\d+(?:,\d+)?%|-)\s+"
    r"(?P<icms>-?\d+(?:\.\d{3})*,\d{2}|-)\s+"
    r"(?P<value>-?\d+(?:\.\d{3})*,\d{2})$"
)

# --- Itens Eventuais (layout: QTY DESC PARCELA FRANQUIA CONSUMO PERIODO DIAS VALOR) ---
# Com consumo real (ex: 19,77GB) ou consumo zerado ("0")
RX_EVENTUAIS = re.compile(
    r"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)\s+-\s+-\s+"
    r"(?P<consumption>(?:\S+(?:GB|MB|KB)|0))\s+-\s+-\s+"
    r"(?P<value>-?\d+,\d{2})\s*$"
)
# Sem consumo (todos "-")
RX_EVENTUAIS_NO_CONS = re.compile(
    r"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)"
    r"(?:\s+-){5}\s+"
    r"(?P<value>-?\d+,\d{2})\s*$"
)

# --- SVA Detalhe Total (layout: # DATA HORA ORIGEM DESC NUMERO TIPO PACOTE - - VALOR) ---
# Exemplo: 1 09/04/25 - 02:54:18 RJ AREA 21 TIM Saude Mensal 00700001003511 N FP - - 14,99
RX_SVA_DETAIL = re.compile(
    r"^\s*(?P<seq>\d+)\s+"
    r"(?P<date>\d{2}/\d{2}/\d{2})\s+-\s+(?P<time>\d{2}:\d{2}:\d{2})\s+"
    r"\S+\s+AREA\s+\d{2}\s+"           # origem (ex: RJ AREA 21)
    r"(?P<desc>.+?)\s+"
    r"(?P<number>\d{10,})\s+"
    r"[A-Z/]+\s+[A-Z]+\s+"
    r"-\s+-\s+"
    r"(?P<value>-?\d+,\d{2})\s*$"
)
# Linha de total SVA (ex: "3 - 2 2 29,98")
RX_SVA_SUMMARY = re.compile(
    r"^\s*(?P<seq>\d+)\s+-\s+\d+\s+\d+\s+(?P<value>-?\d+,\d{2})\s*$"
)

# --- Informações Complementares (layout: QTY DESC PARCELA PERÍODO DIAS VALOR) ---
RX_INFO_COMPL = re.compile(
    r"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)\s+"
    r"(?P<parcel>(?:\d+/\d+|-))\s+"
    r"(?P<period>\d{2}/\d{2}\s+a\s+\d{2}/\d{2})\s+"
    r"(?P<days>\d+)\s+"
    r"(?P<value>-?\d+,\d{2})\s*$"
)
# --- Chamadas resumo (layout: QTY DESC PARCELA FRANQUIA CONSUMO PERIODO DIAS VALOR) ---
RX_CHAMADAS = re.compile(
    r"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)\s+-\s+-\s+"
    r"(?P<consumption>\d{1,3}m\d{2}s)\s+-\s+-\s+"
    r"(?P<value>-?\d+,\d{2})\s*$"
)

# --- Mensalidades Adicionais (layout: QTY DESC PARCELA FRANQUIA CONSUMO PERIODO DIAS VALOR) ---
# Ex.: "1 Apple Music SVA Mes - - 0 - - 21,40" (consumo "0" e sem período)
#      "1 Internet 30GB - - - 25/04 a 24/05 30 0,00"
RX_MENSALIDADE = re.compile(
    rf"^\s*(?P<qty>\d+)\s+(?P<desc>.+?)\s+{RX_PARCEL}\s+"
    rf"(?P<franchise>Ilimitado|-)\s+(?P<consumption>\S+)\s+"
    rf"{RX_PERIOD}\s+{RX_DAYS}\s+{RX_VALUE}\s*$"
)

###############################################################################
# Parser                                                                      #
###############################################################################
class TimBillParser:
    """Extrai e estrutura faturas TIM."""

    def __init__(self, *, x_tolerance: float = 1.5, y_tolerance: float = 3.0):
        self.x_tol = x_tolerance
        self.y_tol = y_tolerance
        self.data: Dict[str, pd.DataFrame] = {}

    # ---------------- API pública ----------------
    def parse_pdf(self, pdf_bytes: io.BytesIO) -> Dict[str, pd.DataFrame]:

        # 1. Extrai texto bruto de todas as páginas
        with pdfplumber.open(pdf_bytes) as pdf:
            raw_text = "\n".join(
                page.extract_text(x_tolerance=self.x_tol,
                                y_tolerance=self.y_tol) or ""
                for page in pdf.pages
            )

        # ------------------------------------------------------------------
        # 2. Captura o bloco FATURA RESUMO inteiro
        bloco_pat = re.compile(
            r"FATURA\s+RESUMO(?P<body>.*?)Total\s+geral\s+R\$\s*(?P<total>-?[\d\.,]+)",
            re.S | re.I,
        )
        resumo_items = []
        resumo_metadata = self._extract_fatura_metadata(raw_text)
        if resumo_metadata.get("period"):
            resumo_items.append(
                dict(qty=None, desc="PERÍODO", parcel=None, period=resumo_metadata["period"],
                    days=None, value=None, franchise=None, consumption=None, msisdn=None,
                    emissao=None, section_total=None)
            )
        if resumo_metadata.get("emissao"):
            resumo_items.append(
                dict(qty=None, desc="EMISSÃO", parcel=None, period=None,
                    days=None, value=None, franchise=None, consumption=None, msisdn=None,
                    emissao=resumo_metadata["emissao"], section_total=None)
            )
        m_resumo = bloco_pat.search(raw_text)
        if m_resumo:
            corpo = m_resumo.group("body")
            total_val = float(m_resumo.group("total").replace(".", "").replace(",", "."))
            for ln in corpo.splitlines():
                ln = _normalize_line(ln)
                if not ln:
                    continue
                m_ln = re.match(r"(.+?)\s+R\$\s*([-\d\.,]+)", ln)
                if m_ln:
                    resumo_items.append(
                        dict(qty=None, desc=m_ln.group(1), parcel=None, period=None,
                            days=None,
                            value=float(m_ln.group(2).replace(".", "").replace(",", ".")),
                            franchise=None, consumption=None, msisdn=None,
                            section_total=None)
                    )
            resumo_items.append(
                dict(qty=None, desc="Total geral", parcel=None, period=None,
                    days=None, value=total_val, franchise=None, consumption=None,
                    msisdn=None, section_total=None)
            )
            # 3. Remove o bloco para que não polua a etapa linha-a-linha
            raw_text = raw_text.replace(m_resumo.group(0), "")

        # ------------------------------------------------------------------
        # 4. Processa o restante normalmente
        dfs = self._parse_text(raw_text)
        if resumo_items:
            dfs["Fatura Resumo"] = pd.DataFrame(resumo_items)

        self.data = dfs
        return dfs

    # ---------------- interno -------------------
    def _extract_fatura_metadata(self, text: str) -> dict[str, str]:
        normalized_lines = [_normalize_line(line) for line in text.splitlines()]
        normalized_text = " ".join(line for line in normalized_lines if line)
        if m := RX_FATURA_METADATA.search(normalized_text):
            return {
                "period": m.group("period"),
                "emissao": m.group("emissao"),
            }
        return {}

    def _parse_text(self, text: str) -> Dict[str, pd.DataFrame]:
        """Quebra o texto extraído em DataFrames por seção."""
        buf: Dict[str, list] = {k: [] for k in SECTION_HEADERS if not k.startswith("IGNORE")}
        current_sec = current_msisdn = None
        current_total: float | None = None
        last_sva_detail_item: dict | None = None

        for raw in text.splitlines():
            line = _normalize_line(raw)
            if not line:
                continue

            # cabeçalho de MSISDN (3 variantes)
            for rx_ms in (RX_MSISDN_HEADER, RX_MSISDN_SEU_NUM, RX_MSISDN_DETALHE):
                if (m := rx_ms.search(line)):
                    current_msisdn = m.group("msisdn").replace(" ", "")
                    break
            else:
                m = None
            if m:
                continue

            # section break — encerra seção atual sem iniciar outra
            if RX_SECTION_BREAK.match(line):
                current_sec = None
                continue

            # detecta novo cabeçalho (a menos que estejamos em Fatura Resumo)
            if current_sec != "Fatura Resumo":
                sec = next((s for s, rx in SECTION_REGEX.items() if rx.match(line)), None)
            else:
                sec = None

            if (sec in ("IGNORE Ilimitados", "IGNORE Detalhamento")):
                current_sec = None
                continue
            if sec is not None:
                current_sec = sec
                last_sva_detail_item = None
                mt = RX_SECTION_TOTAL.search(line)
                current_total = float(mt.group(1).replace(".", "").replace(",", ".")) if mt else None
                continue

            if not current_sec:
                if line[0].isdigit():
                    item = self._parse_sva_detail(line)
                    if item:
                        current_sec = "SVA Detalhe Total"
                        current_total = None
                        item["msisdn"] = current_msisdn
                        item["section_total"] = None
                        buf.setdefault(current_sec, []).append(item)
                        last_sva_detail_item = item
                continue

            # pula cabeçalhos de página
            if re.match(r"^Página\s+\d+\s+de\s+\d+", line, re.I):
                continue

            if current_sec == "DANFE-COM":
                item = self._parse_danfe(line)
                if item:
                    item["msisdn"] = None
                    item["section_total"] = current_total
                    buf.setdefault(current_sec, []).append(item)
                    if item.get("is_total"):
                        current_sec = None
                continue

            if current_sec == "SVA Detalhe Total":
                if self._is_sva_header_line(line):
                    continue
                if RX_SVA_SUMMARY.match(line):
                    last_sva_detail_item = None
                    continue
                if line[0].isdigit():
                    item = self._parse_sva_detail(line)
                    if item:
                        item["msisdn"] = current_msisdn
                        item["section_total"] = current_total
                        buf.setdefault(current_sec, []).append(item)
                        last_sva_detail_item = item
                    continue
                if last_sva_detail_item:
                    last_sva_detail_item["desc"] = (
                        f"{last_sva_detail_item['desc']} {line}"
                    ).strip()
                continue

            # -------- linhas padrão --------
            if not (line[0].isdigit() or line.lower().startswith("subtotal")):
                continue

            item = self._parse_item_line(line, current_sec)
            if item:
                item["msisdn"] = current_msisdn
                item["section_total"] = current_total
                buf.setdefault(current_sec, []).append(item)

        return {s: pd.DataFrame(lst) for s, lst in buf.items() if lst}

    # ------------------------------------------------------------------
    def _parse_item_line(self, line: str, section: str = "") -> dict | None:
        """Parseia uma linha de item de acordo com a seção."""

        # Ignora subtotal
        if RX_SUBTOTAL.match(line):
            return None

        # ── Mensalidades Adicionais ──
        if section == "Mensalidades Adicionais":
            return self._parse_mensalidades(line)

        # ── Itens Eventuais ──
        if section == "Itens Eventuais":
            return self._parse_eventuais(line)

        # ── SVA Detalhe Total ──
        if section == "SVA Detalhe Total":
            return self._parse_sva_detail(line)

        # ── DANFE-COM ──
        if section == "DANFE-COM":
            return self._parse_danfe(line)

        # ── Chamadas Rede TIM / Chamadas Fora Rede TIM ──
        if section.startswith("Chamadas"):
            return self._parse_chamadas(line)

        # ── Plano / Franquia / SVA / Desconto Franquia / Desconto SVA (layout padrão) ──
        return self._parse_standard(line)

    # ------------------------------------------------------------------
    def _parse_danfe(self, line: str) -> dict | None:
        """DANFE-COM: linhas da tabela de itens e total geral."""
        if line.upper().startswith("ITENS "):
            return None

        if (m := RX_DANFE_TOTAL.match(line)):
            return {
                "desc": "Total geral",
                "unit": None,
                "qty": None,
                "preco_unit": None,
                "pis_cofins": None,
                "bc_icms": None,
                "aliq_icms": None,
                "icms": None,
                "value": _parse_money(m.group("value")),
                "is_total": True,
            }

        if (m := RX_DANFE_ROW.match(line)):
            d = m.groupdict()
            return {
                "desc": d["desc"].strip(),
                "unit": d["unit"],
                "qty": float(d["qty"].replace(",", ".")),
                "preco_unit": _parse_money(d["preco_unit"]),
                "pis_cofins": _parse_money(d["pis_cofins"]),
                "bc_icms": _parse_money(d["bc_icms"]),
                "aliq_icms": None if d["aliq_icms"] == "-" else d["aliq_icms"],
                "icms": _parse_money(d["icms"]),
                "value": _parse_money(d["value"]),
                "is_total": False,
            }

        return None

    # ------------------------------------------------------------------
    def _parse_standard(self, line: str) -> dict | None:
        """Regex padrão: QTY DESC PARCELA [FRANCHISE CONSUMPTION] PERIOD DAYS VALUE."""
        for rx in (RX_CONSUMPTION, RX_SIMPLE):
            if (m := rx.match(line)):
                d = m.groupdict()

                # --- corrige parcel dentro de desc (caso d["parcel"] == "-") ----
                if d["parcel"] == "-":
                    tail = re.search(r"\b(\d+/\d+)$", d["desc"])
                    if tail:
                        d["parcel"] = tail.group(1)
                        d["desc"]   = d["desc"][: tail.start()].rstrip(" -")
                # ----------------------------------------------------------------

                # limpa trailing dashes e valores de franquia residuais do desc
                d["desc"] = re.sub(r"(\s+-\s+\d+(?:GB|MB|KB))(?:\s+-)*\s*$|(?:\s+-)+\s*$", "", d["desc"]).strip()

                value = d["value"]
                d["qty"]   = float(d["qty"])
                d["days"]  = None if d.get("days") in (None, "-") else int(d["days"])
                d["_is_included_value"] = value == "Incluído"
                d["value"] = 0.0 if d["_is_included_value"] else float(value.replace(",", "."))
                return d
        return None

    # ------------------------------------------------------------------
    def _parse_eventuais(self, line: str) -> dict | None:
        """Itens Eventuais: QTY DESC PARCELA FRANQUIA CONSUMO PERIODO DIAS VALUE."""
        # Tenta com consumo real (ex: 19,77GB)
        if (m := RX_EVENTUAIS.match(line)):
            return {
                "qty": float(m.group("qty")),
                "desc": m.group("desc").strip(),
                "parcel": None,
                "period": None,
                "days": None,
                "value": float(m.group("value").replace(",", ".")),
                "franchise": None,
                "consumption": None if m.group("consumption") == "0" else m.group("consumption"),
            }
        # Tenta sem consumo (todos "-")
        if (m := RX_EVENTUAIS_NO_CONS.match(line)):
            return {
                "qty": float(m.group("qty")),
                "desc": m.group("desc").strip(),
                "parcel": None,
                "period": None,
                "days": None,
                "value": float(m.group("value").replace(",", ".")),
                "franchise": None,
                "consumption": None,
            }
        return None

    # ------------------------------------------------------------------
    def _parse_mensalidades(self, line: str) -> dict | None:
        """Mensalidades Adicionais: QTY DESC PARCELA FRANQUIA CONSUMO PERIODO DIAS VALOR."""
        if (m := RX_MENSALIDADE.match(line)):
            d = m.groupdict()
            d["qty"] = float(d["qty"])
            d["desc"] = d["desc"].strip()
            d["parcel"] = None if d["parcel"] == "-" else d["parcel"]
            d["franchise"] = None if d["franchise"] == "-" else d["franchise"]
            d["consumption"] = None if d["consumption"] in ("-", "0") else d["consumption"]
            d["period"] = None if d["period"] == "-" else d["period"]
            d["days"] = None if d["days"] == "-" else int(d["days"])
            value = d["value"]
            d["_is_included_value"] = value == "Incluído"
            d["value"] = 0.0 if d["_is_included_value"] else float(value.replace(",", "."))
            return d
        # Layout de 6 colunas (sem franquia/consumo) → regex padrão
        return self._parse_standard(line)

    # ------------------------------------------------------------------
    def _is_sva_header_line(self, line: str) -> bool:
        return bool(
            re.match(r"^(DURAÇÃO/VOLUME|#\s+DATA\s*/\s*HORA)\b", line, re.I)
        )

    # ------------------------------------------------------------------
    def _parse_sva_detail(self, line: str) -> dict | None:
        """SVA Detalhe Total: linhas com data/hora e linha de totalização."""
        # Linha de detalhe com data/hora
        if (m := RX_SVA_DETAIL.match(line)):
            return {
                "qty": 1.0,
                "desc": m.group("desc").strip(),
                "parcel": None,
                "period": m.group("date"),
                "days": None,
                "value": float(m.group("value").replace(",", ".")),
                "franchise": None,
                "consumption": None,
            }
        return None

    # ------------------------------------------------------------------
    def _parse_chamadas(self, line: str) -> dict | None:
        """Chamadas Rede TIM / Fora Rede: QTY DESC - - CONSUMPTION - - VALUE."""
        if (m := RX_CHAMADAS.match(line)):
            return {
                "qty": float(m.group("qty")),
                "desc": m.group("desc").strip(),
                "parcel": None,
                "period": None,
                "days": None,
                "value": float(m.group("value").replace(",", ".")),
                "franchise": None,
                "consumption": m.group("consumption"),
            }
        # Fallback para o padrão (algumas chamadas usam formato padrão)
        return self._parse_standard(line)