from __future__ import annotations

from pathlib import Path

from django.conf import settings

from review_agent.models import FileSummaryBatch
from review_agent.regulatory_review.services.llm_review import review_condition_fields
from review_agent.regulatory_review.services.text_extract import extract_text


OPTION_FIELDS = {
    "product_category": ["体外诊断试剂", "医疗器械", "其他"],
    "registration_type": ["首次注册", "变更注册", "延续注册"],
    "clinical_evaluation_path": ["临床试验", "免临床", "同品种比对", "待确认"],
}


def detect_regulatory_condition_candidates(summary_batch: FileSummaryBatch) -> dict[str, dict[str, object]]:
    """Infers review-scope conditions from the summary batch and file names."""

    corpus_parts = [summary_batch.product_name or ""]
    field_candidates: dict[str, str] = {}
    field_sources: dict[str, str] = {}
    for item in summary_batch.items.order_by("file_index"):
        corpus_parts.extend([item.directory_level, item.file_name, item.relative_path])
        review = _extract_item_fields(item)
        extracted = review.get("selected_fields", {})
        sources = review.get("selected_sources", {})
        field_candidates.update({key: value for key, value in extracted.items() if value and key not in field_candidates})
        field_sources.update({key: value for key, value in sources.items() if value and key not in field_sources})
        corpus_parts.extend(extracted.values())
    corpus = "\n".join(part for part in corpus_parts if part)
    product_name = field_candidates.get("产品名称") or _safe_summary_product_name(summary_batch.product_name)

    return {
        "product_category": {
            "label": "产品类别",
            "input_type": "select",
            "options": OPTION_FIELDS["product_category"],
            "suggested": _detect_product_category(corpus),
        },
        "registration_type": {
            "label": "注册类型",
            "input_type": "select",
            "options": OPTION_FIELDS["registration_type"],
            "suggested": _detect_registration_type(corpus),
        },
        "clinical_evaluation_path": {
            "label": "临床评价路径",
            "input_type": "select",
            "options": OPTION_FIELDS["clinical_evaluation_path"],
            "suggested": _detect_clinical_path(corpus),
        },
        "product_name": {
            "label": "产品名称",
            "input_type": "text",
            "suggested": product_name,
            "source": field_sources.get("产品名称", "summary" if product_name else ""),
        },
        "model_spec": {
            "label": "型号规格",
            "input_type": "text",
            "suggested": field_candidates.get("型号规格", ""),
            "source": field_sources.get("型号规格", ""),
        },
        "intended_use": {
            "label": "预期用途",
            "input_type": "text",
            "suggested": field_candidates.get("预期用途", ""),
            "source": field_sources.get("预期用途", ""),
        },
    }


def _extract_item_fields(item) -> dict[str, object]:
    path = Path(item.storage_path)
    if not path.is_absolute():
        path = Path(settings.MEDIA_ROOT) / item.storage_path
    if not path.exists():
        return {}
    result = extract_text(path)
    if result.status != "success" or not result.field_candidates:
        return {}
    return review_condition_fields(
        text=result.front_text or result.text,
        rule_fields=result.field_candidates,
        file_context=f"{item.directory_level}\n{item.file_name}\n{item.relative_path}",
    )


def _safe_summary_product_name(product_name: str) -> str:
    value = (product_name or "").strip()
    if not value:
        return ""
    if any(keyword in value for keyword in ["第1章", "第2章", "监管信息", "综述资料", "非临床资料", "章节目录"]):
        return ""
    return value


def _detect_product_category(corpus: str) -> str:
    if any(keyword in corpus for keyword in ["体外诊断", "检测试剂", "试剂盒", "IVD"]):
        return "体外诊断试剂"
    if "医疗器械" in corpus:
        return "医疗器械"
    return "其他"


def _detect_registration_type(corpus: str) -> str:
    if "延续" in corpus:
        return "延续注册"
    if "变更" in corpus:
        return "变更注册"
    return "首次注册"


def _detect_clinical_path(corpus: str) -> str:
    if "免临床" in corpus or "免于临床" in corpus:
        return "免临床"
    if "同品种" in corpus or "同类" in corpus:
        return "同品种比对"
    if "临床试验" in corpus:
        return "临床试验"
    return "待确认"