backend/core/projects/data_completeness.py

"""
Pre-generation data completeness gate.

Before full autopilot: ready_to_generate | blocked_missing_fields |
blocked_dossier | structure_only.
"""
from __future__ import annotations

from typing import Any, Dict, List, Optional, Tuple


STATUS_READY = "ready_to_generate"
STATUS_MISSING = "blocked_missing_fields"
STATUS_DOSSIER = "blocked_dossier"
STATUS_STRUCTURE = "structure_only"
STATUS_BLOCKED_CONSENT = "blocked_consent"


def _get_nested(d: Dict[str, Any], *keys: str) -> Any:
    cur: Any = d
    for k in keys:
        if not isinstance(cur, dict):
            return None
        cur = cur.get(k)
    return cur


def _has_value(v: Any) -> bool:
    if v is None:
        return False
    if isinstance(v, str):
        s = v.strip()
        if not s or s.startswith("[DO WERYFIKACJI") or s.lower() in ("brak", "n/a", "tbd"):
            return False
        return True
    if isinstance(v, (list, dict)):
        return len(v) > 0
    return True


def extract_company_field_map(external_context: Optional[Dict[str, Any]]) -> Dict[str, Any]:
    ext = dict(external_context or {})
    company = ext.get("company_data") if isinstance(ext.get("company_data"), dict) else {}
    return {
        "nip": company.get("nip") or ext.get("nip"),
        "company_name": company.get("name") or company.get("company_name") or ext.get("company_name"),
        "pkd_codes": company.get("pkd")
        or company.get("pkd_codes")
        or company.get("pkd_list")
        or ext.get("pkd_codes"),
        "msp_status": company.get("msp_status")
        or company.get("sme_status")
        or ext.get("msp_status")
        or _get_nested(ext, "msp_analysis", "status"),
        "de_minimis": company.get("de_minimis")
        or ext.get("de_minimis")
        or company.get("de_minimis_eur"),
        "closed_financial_year": company.get("closed_financial_year")
        or ext.get("closed_financial_year")
        or company.get("financial_year_closed"),
        "revenue_3y": company.get("revenue_3y")
        or company.get("revenues")
        or ext.get("revenue_3y"),
        "employment_fte": company.get("employment_fte")
        or company.get("fte")
        or company.get("employees")
        or ext.get("employment_fte"),
        "address": company.get("address") or company.get("adres"),
        "regon": company.get("regon") or ext.get("regon"),
    }


def extract_project_field_map(
    external_context: Optional[Dict[str, Any]],
    *,
    description: str = "",
    title: str = "",
) -> Dict[str, Any]:
    ext = dict(external_context or {})
    proj = ext.get("project_facts") if isinstance(ext.get("project_facts"), dict) else {}
    return {
        "project_description": description or proj.get("description") or ext.get("project_description"),
        "innovation_description": proj.get("innovation") or ext.get("innovation_description"),
        "eurogrant_programme": proj.get("eurogrant_programme") or ext.get("eurogrant_programme"),
        "eurogrant_call": proj.get("eurogrant_call") or ext.get("eurogrant_call"),
        "eurogrant_role": proj.get("eurogrant_role") or ext.get("eurogrant_role"),
        "specialist_analysis_yes_no": proj.get("specialist_analysis_yes_no")
        or ext.get("specialist_analysis_yes_no"),
        "lump_sum_amount_pln": proj.get("lump_sum_amount_pln") or ext.get("lump_sum_amount_pln"),
        "title": title or ext.get("title"),
    }


_FIELD_LABELS_PL = {
    "nip": "NIP firmy",
    "company_name": "Nazwa firmy",
    "pkd_codes": "Kody PKD (lista)",
    "msp_status": "Status MŚP",
    "de_minimis": "Pomoc de minimis (wykorzystana)",
    "closed_financial_year": "Potwierdzenie zamkniętego roku obrotowego (≥12 mies.)",
    "revenue_3y": "Przychody za ostatnie 3 lata",
    "employment_fte": "Zatrudnienie FTE",
    "project_description": "Opis projektu",
    "innovation_description": "Opis innowacji / TRL",
    "eurogrant_programme": "Program UE (target Eurograntu)",
    "eurogrant_call": "Konkurs / call / Work Programme",
    "eurogrant_role": "Rola (samodzielny / koordynator / członek)",
    "specialist_analysis_yes_no": "Czy analizy specjalistyczne (tak/nie + uzasadnienie)",
    "lump_sum_amount_pln": "Kwota ryczałtowa (PLN)",
}


def evaluate_data_completeness(
    *,
    external_context: Optional[Dict[str, Any]] = None,
    instrument_schema: Optional[Dict[str, Any]] = None,
    description: str = "",
    title: str = "",
    grounding_mode: str = "",
    dossier_level: str = "",
) -> Dict[str, Any]:
    """
    Returns readiness status + missing field descriptors for HITL UI.
    """
    ext = dict(external_context or {})
    schema = instrument_schema or ext.get("instrument_schema") or {}
    mode = (grounding_mode or ext.get("grounding_mode") or "").lower()
    level = (
        dossier_level
        or (ext.get("program_dossier") or {}).get("readiness", {}).get("level")
        or ext.get("dossier_readiness")
        or ""
    )
    level = str(level).lower()

    if mode == "structure_only":
        # Structure-only consent allows *structural* generation without a regulation
        # pack — but Full Autopilot must still pass the eligibility spine (MŚP /
        # de minimis). Early-return used to skip apply_eligibility_to_completeness.
        result = {
            "status": STATUS_STRUCTURE,
            "ready_to_generate": True,  # structural only — allowed with consent
            "full_autopilot_allowed": True,  # may be cleared by spine below
            "missing_fields": [],
            "warnings": [
                "Tryb structuralny: generacja bez ugruntowania w regulaminie — nie składać bez weryfikacji."
            ],
            "instrument_family": schema.get("family"),
            "dossier_level": level or "blind",
        }
        try:
            from core.eligibility.spine import (
                apply_eligibility_to_completeness,
                build_eligibility_report,
            )

            company = extract_company_field_map(ext)
            report = build_eligibility_report(
                nip=str(company.get("nip") or ""),
                company_data=ext.get("company_data")
                if isinstance(ext.get("company_data"), dict)
                else {},
                external_context=ext,
            )
            result = apply_eligibility_to_completeness(result, report)
            # Preserve structure_only generation consent; spine still owns autopilot.
            result["ready_to_generate"] = True
            if result.get("status") != STATUS_STRUCTURE:
                result["status"] = STATUS_STRUCTURE
            warns = list(result.get("warnings") or [])
            structural_warn = (
                "Tryb structuralny: generacja bez ugruntowania w regulaminie — nie składać bez weryfikacji."
            )
            if structural_warn not in warns:
                warns.insert(0, structural_warn)
            result["warnings"] = warns
        except Exception:
            pass
        return result

    if mode == "blocked" or (level == "blind" and mode != "structure_only"):
        # Dossier blind without consent
        from core.projects.generation_consent import (
            resolve_grounding_mode,
            GROUNDING_STRUCTURE,
            GROUNDING_REGULATION,
        )

        gm = resolve_grounding_mode(ext)
        if gm == GROUNDING_STRUCTURE:
            return evaluate_data_completeness(
                external_context={**ext, "grounding_mode": "structure_only"},
                instrument_schema=schema,
                description=description,
                title=title,
                grounding_mode="structure_only",
                dossier_level=level,
            )
        if gm != GROUNDING_REGULATION and level == "blind":
            return {
                "status": STATUS_DOSSIER,
                "ready_to_generate": False,
                "full_autopilot_allowed": False,
                "missing_fields": [
                    {
                        "field": "regulation_pack",
                        "label": "Regulamin / paczka dokumentów naboru",
                        "category": "dossier",
                        "hint": "Dołącz URL/PDF regulaminu albo jawnie zaakceptuj tryb strukturalny.",
                    }
                ],
                "warnings": ["Dossier blind — brak reguł naboru w systemie."],
                "instrument_family": schema.get("family"),
                "dossier_level": level or "blind",
            }

    company = extract_company_field_map(ext)
    project = extract_project_field_map(ext, description=description, title=title)
    required_company = list(schema.get("required_company_fields") or ["nip", "company_name"])
    required_project = list(schema.get("required_project_fields") or ["project_description"])

    missing: List[Dict[str, str]] = []
    for f in required_company:
        if not _has_value(company.get(f)):
            missing.append(
                {
                    "field": f,
                    "label": _FIELD_LABELS_PL.get(f, f),
                    "category": "company",
                    "hint": "Uzupełnij w profilu firmy lub wgraj dokumenty (CEIDG, sprawozdania).",
                }
            )
    for f in required_project:
        if not _has_value(project.get(f)):
            missing.append(
                {
                    "field": f,
                    "label": _FIELD_LABELS_PL.get(f, f),
                    "category": "project",
                    "hint": "Uzupełnij w opisie projektu / faktach projektu (project_facts).",
                }
            )

    # Minimum always: NIP + name for full autopilot quality
    if not _has_value(company.get("nip")):
        if not any(m["field"] == "nip" for m in missing):
            missing.append(
                {
                    "field": "nip",
                    "label": _FIELD_LABELS_PL["nip"],
                    "category": "company",
                    "hint": "Wymagane do full autopilot.",
                }
            )

    if missing:
        result = {
            "status": STATUS_MISSING,
            "ready_to_generate": False,
            "full_autopilot_allowed": False,
            "missing_fields": missing,
            "warnings": [],
            "instrument_family": schema.get("family"),
            "dossier_level": level or "unknown",
            "present_company": {k: _has_value(v) for k, v in company.items()},
            "present_project": {k: _has_value(v) for k, v in project.items()},
        }
    else:
        result = {
            "status": STATUS_READY,
            "ready_to_generate": True,
            "full_autopilot_allowed": True,
            "missing_fields": [],
            "warnings": [],
            "instrument_family": schema.get("family"),
            "dossier_level": level or "unknown",
            "present_company": {k: _has_value(v) for k, v in company.items()},
            "present_project": {k: _has_value(v) for k, v in project.items()},
        }

    # Eligibility spine 2026 (MŚP + de minimis) — product gate when flag on
    try:
        from core.eligibility.spine import (
            apply_eligibility_to_completeness,
            build_eligibility_report,
        )

        report = build_eligibility_report(
            nip=str(company.get("nip") or ""),
            company_data=ext.get("company_data") if isinstance(ext.get("company_data"), dict) else {},
            external_context=ext,
        )
        result = apply_eligibility_to_completeness(result, report)
    except Exception:
        pass
    return result


def apply_project_facts(
    external_context: Optional[Dict[str, Any]],
    facts: Dict[str, Any],
) -> Dict[str, Any]:
    """Merge user-provided facts into external_context.project_facts + top-level mirrors."""
    ext = dict(external_context or {})
    pf = dict(ext.get("project_facts") or {}) if isinstance(ext.get("project_facts"), dict) else {}
    for k, v in (facts or {}).items():
        if v is None:
            continue
        pf[k] = v
        # mirror common fields to top-level for extractors
        if k in (
            "eurogrant_programme",
            "eurogrant_call",
            "eurogrant_role",
            "specialist_analysis_yes_no",
            "lump_sum_amount_pln",
        ):
            ext[k] = v
    ext["project_facts"] = pf
    # optional company patch
    company_patch = facts.get("company") if isinstance(facts.get("company"), dict) else None
    if company_patch:
        cd = dict(ext.get("company_data") or {}) if isinstance(ext.get("company_data"), dict) else {}
        cd.update({k: v for k, v in company_patch.items() if v is not None})
        ext["company_data"] = cd
    return ext