src/Matching Functions/contract_match.py

from __future__ import annotations

from dataclasses import dataclass, field
from datetime import date, datetime
from decimal import Decimal, ROUND_HALF_EVEN, getcontext
from typing import Any, Callable, Optional

import re
import unicodedata

from rapidfuzz import fuzz

from two_way_match import MatchExceptionRecord, InvoiceLine, to_decimal

# ============================================================
# Decimal hygiene
# ============================================================
getcontext().prec = 28
getcontext().rounding = ROUND_HALF_EVEN

D0 = Decimal("0")
D100 = Decimal("100")


# ============================================================
# Data models
# ============================================================
@dataclass(frozen=True)
class RateCardItem:
    id: int
    description: Optional[str] = None
    description_canonical: Optional[str] = None
    unit_price: Optional[Decimal] = None
    item_type: Optional[str] = None  # goods | services | None


@dataclass(frozen=True)
class ContractRecord:
    id: int
    tenant_id: int
    vendor_id: int
    entity_id: int
    currency: str
    status: str = "active"
    deleted_at: Any = None
    effective_from: Optional[date] = None
    effective_to: Optional[date] = None
    created_at: Optional[datetime] = None
    total_value: Optional[Decimal] = None
    cumulative_billed_amount: Optional[Decimal] = None
    rate_card_items: Optional[list[RateCardItem]] = None


@dataclass(frozen=True)
class ContractInvoiceHeader:
    id: int
    tenant_id: int
    entity_id: int
    resolved_vendor_id: Optional[int]
    currency: Optional[str]
    invoice_date: Optional[date]
    service_start_date: Optional[date] = None
    service_end_date: Optional[date] = None
    total_amount: Optional[Decimal] = None
    advisory_flags: tuple[str, ...] = ()


@dataclass
class ContractLineResult:
    invoice_line_id: int
    rate_card_item_id: Optional[int]
    line_status: str = "matched"   # matched | no_match | exception
    line_zone: str = "green"       # green | amber | red
    price_var_pct: Optional[Decimal] = None
    allocated_inv_amount: Optional[Decimal] = None
    allocated_rate_amount: Optional[Decimal] = None
    exception_type: Optional[str] = None
    exception_detail: dict[str, Any] = field(default_factory=dict)


@dataclass
class ContractMatchResult:
    match_type: str = "contract"
    overall_status: str = "exception"  # matched | partial_match | exception
    match_zone: str = "red"            # green | amber | red | critical
    documents_status: str = "pending_approval"

    lines_total: int = 0
    lines_matched: int = 0
    lines_in_exception: int = 0

    advisory_flags: list[str] = field(default_factory=list)
    risk_flags: list[str] = field(default_factory=list)
    skipped_steps: list[dict] = field(default_factory=list)
    header_exceptions: list[MatchExceptionRecord] = field(default_factory=list)
    match_exceptions: list[MatchExceptionRecord] = field(default_factory=list)
    match_line_results: list[ContractLineResult] = field(default_factory=list)

    selected_contract_id: Optional[int] = None


@dataclass(frozen=True)
class ContractConfig:
    # RapidFuzz thresholds for rate-card / tiebreaker matching
    high_threshold: float = 90.0
    medium_threshold: float = 80.0
    score_cutoff: float = 80.0

    # Price variance bands (same idea as 2-way; tune to your config table)
    price_goods_green: Decimal = Decimal("5")
    price_goods_amber: Decimal = Decimal("10")
    price_services_green: Decimal = Decimal("10")
    price_services_amber: Decimal = Decimal("15")

    precision: int = 28


# ============================================================
# Helpers
# ============================================================
def canonicalise_identifier(value: Optional[str]) -> Optional[str]:
    if value is None:
        return None
    value = unicodedata.normalize("NFKC", str(value)).upper()
    value = re.sub(r"[^A-Z0-9]+", "", value)
    return value or None


def norm_currency(value: Optional[str]) -> Optional[str]:
    return value.upper().strip() if value else None


def price_variance_pct(inv_price: Optional[Decimal], ref_price: Optional[Decimal]) -> Optional[Decimal]:
    if inv_price is None or ref_price is None or ref_price == 0:
        return None
    return (abs(inv_price - ref_price) / ref_price * D100).quantize(Decimal("0.0001"))


def _bands(item_type: Optional[str], cfg: ContractConfig) -> tuple[Decimal, Decimal]:
    item_type = (item_type or "services").lower()
    if item_type == "goods":
        return cfg.price_goods_green, cfg.price_goods_amber
    return cfg.price_services_green, cfg.price_services_amber


def _line_zone_from_price(price_pct: Optional[Decimal], item_type: Optional[str], cfg: ContractConfig) -> str:
    if price_pct is None:
        return "green"
    green, amber = _bands(item_type, cfg)
    if price_pct <= green:
        return "green"
    if price_pct <= amber:
        return "amber"
    return "red"


def _is_service_window_contained(
    contract: ContractRecord,
    invoice_start: Optional[date],
    invoice_end: Optional[date],
) -> bool:
    if invoice_start is None or invoice_end is None:
        return False
    if contract.effective_from is None:
        return False
    effective_to = contract.effective_to or date(9999, 12, 31)
    return contract.effective_from <= invoice_start and invoice_end <= effective_to


def _joined_invoice_descriptions(inv_lines: list[InvoiceLine]) -> str:
    parts = []
    for line in inv_lines:
        if line.description_canonical:
            parts.append(line.description_canonical)
        elif line.description:
            parts.append(canonicalise_identifier(line.description) or "")
    return " ".join(p for p in parts if p).strip()


def rapidfuzz_best_match(
    query: str,
    candidates: list[dict[str, Any]],
    *,
    high_threshold: float,
    medium_threshold: float,
    score_cutoff: float,
) -> dict[str, Any]:
    """
    Simple local matcher.
    Swap this out with your platform name-match algorithm later if required.
    """
    if not query or not candidates:
        return {"decision": "unmatched", "score_tier": None, "score": 0.0, "record_id": None}

    best_record_id = None
    best_score = 0.0

    for cand in candidates:
        record_id = cand.get("record_id")
        names = cand.get("names") or []
        for name in names:
            if not name:
                continue
            score = float(fuzz.WRatio(query, name, score_cutoff=score_cutoff))
            if score > best_score:
                best_score = score
                best_record_id = record_id

    if best_record_id is None or best_score < score_cutoff:
        return {"decision": "unmatched", "score_tier": None, "score": 0.0, "record_id": None}

    if best_score >= high_threshold:
        tier = "high"
    elif best_score >= medium_threshold:
        tier = "medium"
    else:
        tier = "low"

    return {
        "decision": "matched" if tier in {"high", "medium"} else "suggestion",
        "score_tier": tier,
        "score": round(best_score, 2),
        "record_id": best_record_id,
    }


def _contract_stub_result(
    invoice: ContractInvoiceHeader,
    inv_lines: list[InvoiceLine],
    exception_type: str,
    detail: dict[str, Any],
) -> ContractMatchResult:
    res = ContractMatchResult(
        overall_status="exception",
        match_zone="red",
        documents_status="pending_approval",
        lines_total=len(inv_lines),
        lines_matched=0,
        lines_in_exception=len(inv_lines),
    )
    res.header_exceptions.append(
        MatchExceptionRecord(
            exception_type=exception_type,
            exception_detail=detail,
        )
    )
    return res


# ============================================================
# Contract candidate selection
# ============================================================
def select_contract_candidate(
    invoice: ContractInvoiceHeader,
    inv_lines: list[InvoiceLine],
    contracts: list[ContractRecord],
    cfg: ContractConfig,
) -> tuple[Optional[ContractRecord], Optional[MatchExceptionRecord]]:
    """
    Returns:
      - selected contract
      - or a MatchExceptionRecord for stub paths
    """
    inv_currency = norm_currency(invoice.currency)

    initial_candidates: list[dict[str, Any]] = []
    for c in contracts:
        if c.deleted_at is not None:
            continue
        if c.status != "active":
            continue
        if c.tenant_id != invoice.tenant_id:
            continue
        if c.vendor_id != invoice.resolved_vendor_id:
            continue
        if c.entity_id != invoice.entity_id:
            continue
        if norm_currency(c.currency) != norm_currency(invoice.currency):
            continue
        if invoice.invoice_date is None:
            continue
        if c.effective_from is None:
            continue
        effective_to = c.effective_to or date(9999, 12, 31)
        if not (c.effective_from <= invoice.invoice_date < effective_to):
            continue

        initial_candidates.append(
            {
                "contract": c,
                "elimination_reason": None,
            }
        )

    if not initial_candidates:
        return None, MatchExceptionRecord(
            exception_type="contract_not_found",
            exception_detail={
                "reason": "initial_query_returned_zero_rows",
                "tenant_id": invoice.tenant_id,
                "vendor_id": invoice.resolved_vendor_id,
                "entity_id": invoice.entity_id,
                "currency": inv_currency,
                "invoice_date": str(invoice.invoice_date) if invoice.invoice_date else None,
            },
        )

    remaining = [x["contract"] for x in initial_candidates]

    # Tiebreaker 1: service-period containment (soft preference — only when both dates present)
    if invoice.service_start_date is not None and invoice.service_end_date is not None:
        contained = [c for c in remaining if _is_service_window_contained(c, invoice.service_start_date, invoice.service_end_date)]
        if contained:
            remaining = contained
        # If none pass containment, keep all and let run_contract_match_core raise billing_period_outside_contract

    # Tiebreaker 2: rate-card best fit via RapidFuzz
    # Query uses joined canonical invoice descriptions.
    joined_desc = _joined_invoice_descriptions(inv_lines)
    if joined_desc:
        scored_candidates: list[dict[str, Any]] = []
        for c in remaining:
            if not c.rate_card_items:
                continue
            item_names = [
                (item.description_canonical or canonicalise_identifier(item.description) or "")
                for item in c.rate_card_items
            ]
            item_names = [n for n in item_names if n]
            if not item_names:
                continue
            scored_candidates.append({"record_id": c.id, "names": item_names})

        if scored_candidates:
            # Contracts without rate card items cannot participate; exclude them now
            # so they don't survive to tiebreaker 3 ahead of contracts that do.
            scored_ids = {x["record_id"] for x in scored_candidates}
            remaining = [c for c in remaining if c.id in scored_ids]

            # Evaluate each contract individually: keep only those whose rate card
            # descriptions yield decision='matched' against the joined invoice text.
            matched_ids = {
                x["record_id"]
                for x in scored_candidates
                if rapidfuzz_best_match(
                    joined_desc,
                    [x],
                    high_threshold=cfg.high_threshold,
                    medium_threshold=cfg.medium_threshold,
                    score_cutoff=cfg.score_cutoff,
                )["decision"] == "matched"
            }
            if matched_ids:
                remaining = [c for c in remaining if c.id in matched_ids]
            else:
                return None, MatchExceptionRecord(
                    exception_type="cannot_match_no_contract",
                    exception_detail={
                        "reason": "rate_card_best_fit_filtered_all_candidates",
                        "joined_invoice_descriptions": joined_desc,
                        "considered_contract_ids": [c.id for c in remaining],
                    },
                )

    # Tiebreaker 3: newest created_at DESC, id DESC
    remaining.sort(
        key=lambda c: (
            c.created_at or datetime.min,
            c.id,
        ),
        reverse=True,
    )

    if len(remaining) == 1:
        return remaining[0], None

    if len(remaining) == 0:
        return None, MatchExceptionRecord(
            exception_type="cannot_match_no_contract",
            exception_detail={
                "reason": "all_candidates_removed_by_tiebreakers",
            },
        )

    return None, MatchExceptionRecord(
        exception_type="ambiguous_active_contract",
        exception_detail={
            "reason": "multiple_contracts_remain_after_tiebreakers",
            "surviving_contract_ids": [c.id for c in remaining],
        },
    )


# ============================================================
# Config builder
# ============================================================
def _cfg_from_contract_dict(d: dict) -> ContractConfig:
    def _d(key: str, default: str) -> Decimal:
        return Decimal(str(d.get(key, default)))
    def _f(key: str, default: float) -> float:
        return float(d.get(key, default))
    def _i(key: str, default: int) -> int:
        return int(d.get(key, default))
    return ContractConfig(
        high_threshold=_f("contract_fuzzy_high_threshold", 90.0),
        medium_threshold=_f("contract_fuzzy_medium_threshold", 80.0),
        score_cutoff=_f("contract_fuzzy_score_cutoff", 80.0),
        price_goods_green=_d("contract_price_goods_green", "5"),
        price_goods_amber=_d("contract_price_goods_amber", "10"),
        price_services_green=_d("contract_price_services_green", "10"),
        price_services_amber=_d("contract_price_services_amber", "15"),
        precision=_i("contract_precision", 28),
    )


# ============================================================
# Main contract match core
# ============================================================
def run_contract_match_core(
    invoice: ContractInvoiceHeader,
    inv_lines: list[InvoiceLine],
    contracts: list[ContractRecord],
    *,
    matching_configs: Optional[dict] = None,
) -> ContractMatchResult:
    """
    Contract match used when po_number_canonical is NULL.

    Returns:
      - a stub exception result if no unique contract can be selected
      - otherwise runs billing-period, rate-card, and budget checks

    Note:
      - cumulative_billed_amount is NOT incremented here
      - this is matching-only, not approval
    """
    cfg = _cfg_from_contract_dict(matching_configs or {})
    getcontext().prec = cfg.precision

    result = ContractMatchResult(
        lines_total=len(inv_lines),
        documents_status="pending_approval",
        match_type="contract",
    )
    result.advisory_flags = list(invoice.advisory_flags)

    if invoice.resolved_vendor_id is None:
        return _contract_stub_result(
            invoice,
            inv_lines,
            "cannot_match_no_vendor",
            {
                "reason": "resolved_vendor_id_is_required",
                "tenant_id": invoice.tenant_id,
                "entity_id": invoice.entity_id,
            },
        )

    selected_contract, stub_exc = select_contract_candidate(invoice, inv_lines, contracts, cfg)
    if stub_exc is not None or selected_contract is None:
        # contract_not_found / cannot_match_no_contract / ambiguous_active_contract
        result.header_exceptions.append(stub_exc or MatchExceptionRecord(
            exception_type="cannot_match_no_contract",
            exception_detail={"reason": "no_selected_contract"},
        ))
        result.overall_status = "exception"
        result.match_zone = "red"
        result.lines_in_exception = len(inv_lines)
        return result

    result.selected_contract_id = selected_contract.id

    # ---------------------------------------------------------
    # Contract checks
    # ---------------------------------------------------------
    # 1) Billing period missing
    if invoice.service_start_date is None or invoice.service_end_date is None:
        result.match_exceptions.append(
            MatchExceptionRecord(
                exception_type="billing_period_unverified",
                exception_detail={
                    "reason": "service_start_date_or_service_end_date_missing",
                    "service_start_date": str(invoice.service_start_date) if invoice.service_start_date else None,
                    "service_end_date": str(invoice.service_end_date) if invoice.service_end_date else None,
                },
            )
        )

    # 2) Billing period outside contract
    if invoice.service_start_date is not None and invoice.service_end_date is not None:
        if not _is_service_window_contained(selected_contract, invoice.service_start_date, invoice.service_end_date):
            result.match_exceptions.append(
                MatchExceptionRecord(
                    exception_type="billing_period_outside_contract",
                    exception_detail={
                        "reason": "invoice_service_window_not_contained_in_contract",
                        "service_start_date": str(invoice.service_start_date),
                        "service_end_date": str(invoice.service_end_date),
                        "contract_effective_from": str(selected_contract.effective_from) if selected_contract.effective_from else None,
                        "contract_effective_to": str(selected_contract.effective_to) if selected_contract.effective_to else None,
                    },
                )
            )

    # 3) Rate card exists
    if not selected_contract.rate_card_items:
        result.match_exceptions.append(
            MatchExceptionRecord(
                exception_type="no_rate_card_configured",
                exception_detail={
                    "reason": "contract_has_no_rate_card_items",
                    "contract_id": selected_contract.id,
                },
            )
        )
        result.overall_status = "exception"
        result.match_zone = "red"
        result.lines_in_exception = len(inv_lines)
        return result

    # 4) Per-line rate check
    rate_card_candidates = [
        {
            "record_id": item.id,
            "names": [item.description_canonical or canonicalise_identifier(item.description) or ""],
        }
        for item in selected_contract.rate_card_items
        if (item.description_canonical or item.description)
    ]

    line_matched_count = 0
    line_exception_count = 0

    for inv_line in inv_lines:
        query = inv_line.description_canonical or canonicalise_identifier(inv_line.description) or ""
        if not query:
            result.match_exceptions.append(
                MatchExceptionRecord(
                    exception_type="no_rate_card_match",
                    line_id=inv_line.id,
                    exception_detail={
                        "reason": "invoice_line_description_missing",
                        "invoice_line_id": inv_line.id,
                    },
                )
            )
            line_exception_count += 1
            continue

        rf = rapidfuzz_best_match(
            query=query,
            candidates=rate_card_candidates,
            high_threshold=cfg.high_threshold,
            medium_threshold=cfg.medium_threshold,
            score_cutoff=cfg.score_cutoff,
        )

        if rf["decision"] != "matched" or rf["record_id"] is None:
            result.match_exceptions.append(
                MatchExceptionRecord(
                    exception_type="no_rate_card_match",
                    line_id=inv_line.id,
                    exception_detail={
                        "reason": "rate_card_name_match_failed",
                        "invoice_line_id": inv_line.id,
                        "query": query,
                    },
                )
            )
            line_exception_count += 1
            continue

        rate_item = next((x for x in selected_contract.rate_card_items if x.id == rf["record_id"]), None)
        if rate_item is None:
            result.match_exceptions.append(
                MatchExceptionRecord(
                    exception_type="no_rate_card_match",
                    line_id=inv_line.id,
                    exception_detail={
                        "reason": "matched_rate_card_item_not_found",
                        "invoice_line_id": inv_line.id,
                        "matched_rate_card_item_id": rf["record_id"],
                    },
                )
            )
            line_exception_count += 1
            continue

        # Compare unit price if present
        if inv_line.unit_price is None:
            if "rate_card_match_no_price_compare" not in result.advisory_flags:
                result.advisory_flags.append("rate_card_match_no_price_compare")
            result.match_line_results.append(
                ContractLineResult(
                    invoice_line_id=inv_line.id,
                    rate_card_item_id=rate_item.id,
                    line_status="matched",
                    line_zone="green",
                    price_var_pct=None,
                    allocated_inv_amount=inv_line.amount,
                    allocated_rate_amount=rate_item.unit_price,
                )
            )
            line_matched_count += 1
            continue

        if rate_item.unit_price is None:
            # Name match succeeded; price comparison is impossible. Treat as an
            # advisory (not a hard exception) because the match itself is valid.
            if "rate_card_match_no_price_compare" not in result.advisory_flags:
                result.advisory_flags.append("rate_card_match_no_price_compare")
            result.match_line_results.append(
                ContractLineResult(
                    invoice_line_id=inv_line.id,
                    rate_card_item_id=rate_item.id,
                    line_status="matched",
                    line_zone="green",
                    price_var_pct=None,
                    allocated_inv_amount=inv_line.amount,
                    allocated_rate_amount=None,
                )
            )
            line_matched_count += 1
            continue

        p_var = price_variance_pct(inv_line.unit_price, rate_item.unit_price)
        zone = _line_zone_from_price(p_var, rate_item.item_type, cfg)

        if zone == "red":
            result.match_exceptions.append(
                MatchExceptionRecord(
                    exception_type="rate_card_price_mismatch",
                    line_id=inv_line.id,
                    po_line_id=rate_item.id,
                    exception_detail={
                        "reason": "price_outside_tolerance",
                        "invoice_line_id": inv_line.id,
                        "rate_card_item_id": rate_item.id,
                        "invoice_unit_price": str(inv_line.unit_price),
                        "rate_card_unit_price": str(rate_item.unit_price),
                        "price_var_pct": str(p_var),
                    },
                )
            )
            result.match_line_results.append(
                ContractLineResult(
                    invoice_line_id=inv_line.id,
                    rate_card_item_id=rate_item.id,
                    line_status="exception",
                    line_zone="red",
                    price_var_pct=p_var,
                    allocated_inv_amount=inv_line.amount,
                    allocated_rate_amount=rate_item.unit_price,
                    exception_type="rate_card_price_mismatch",
                )
            )
            line_exception_count += 1
            continue

        result.match_line_results.append(
            ContractLineResult(
                invoice_line_id=inv_line.id,
                rate_card_item_id=rate_item.id,
                line_status="matched",
                line_zone=zone,
                price_var_pct=p_var,
                allocated_inv_amount=inv_line.amount,
                allocated_rate_amount=rate_item.unit_price,
            )
        )
        line_matched_count += 1

    result.lines_matched = line_matched_count
    result.lines_in_exception = line_exception_count + (len(inv_lines) - line_matched_count - line_exception_count)

    # ---------------------------------------------------------
    # Budget ceiling
    # ---------------------------------------------------------
    inv_currency = norm_currency(invoice.currency)
    contract_currency = norm_currency(selected_contract.currency)

    if contract_currency != inv_currency:
        result.match_exceptions.append(
            MatchExceptionRecord(
                exception_type="contract_currency_mismatch",
                exception_detail={
                    "reason": "contract_currency_changed_or_mismatch",
                    "invoice_currency": inv_currency,
                    "contract_currency": contract_currency,
                    "contract_id": selected_contract.id,
                },
            )
        )
        result.overall_status = "exception"
        result.match_zone = "red"
        return result

    if invoice.total_amount is not None:
        cumulative = selected_contract.cumulative_billed_amount or D0
        total_value = selected_contract.total_value
        if total_value is not None and (cumulative + invoice.total_amount) > total_value:
            result.match_exceptions.append(
                MatchExceptionRecord(
                    exception_type="contract_budget_exceeded",
                    exception_detail={
                        "reason": "budget_ceiling_breached",
                        "contract_id": selected_contract.id,
                        "invoice_total_amount": str(invoice.total_amount),
                        "cumulative_billed_amount": str(cumulative),
                        "contract_total_value": str(total_value),
                    },
                )
            )
            result.overall_status = "exception"
            result.match_zone = "red"
            return result

    # ---------------------------------------------------------
    # Overall status / zone
    # ---------------------------------------------------------
    if any(exc.exception_type in {
        "billing_period_unverified",
        "billing_period_outside_contract",
        "no_rate_card_configured",
        "no_rate_card_match",
        "rate_card_price_mismatch",
        "contract_budget_exceeded",
        "contract_currency_mismatch",
    } for exc in result.match_exceptions):
        # If any hard contract exception exists, keep it red.
        if result.lines_matched > 0 and result.lines_in_exception > 0:
            result.overall_status = "partial_match"
        else:
            result.overall_status = "exception"
        result.match_zone = "red"
    else:
        if result.lines_matched == result.lines_total and result.lines_in_exception == 0:
            result.overall_status = "matched"
            result.match_zone = "amber" if any(r.line_zone == "amber" for r in result.match_line_results) else "green"
        elif result.lines_matched > 0:
            result.overall_status = "partial_match"
            result.match_zone = "amber" if any(r.line_zone == "amber" for r in result.match_line_results) else "green"
        else:
            result.overall_status = "exception"
            result.match_zone = "red"

    return result