group_xinghuo_jinrong/scripts/seed/import_script_templates.py

"""Import script templates from the Sprint 2 Markdown dataset."""

from __future__ import annotations

import argparse
import sys
from dataclasses import dataclass
from datetime import datetime, timezone
from pathlib import Path

ROOT = Path(__file__).resolve().parents[2]
sys.path.insert(0, str(ROOT))

from app.advisor_db import AgentSessionLocal
from app.model.entities_advisor import ScriptTemplate
from app.model.advisor_schemas import ComplianceCheckRequest
from app.service.compliance_check_service import ComplianceCheckService

DEFAULT_DATASET = ROOT / "docs" / "开发文档" / "29-Sprint2首批话术模板数据集.md"
ACTOR_ID = "seed:template_test_data"
REQUIRED_HEADERS = [
    "序号",
    "scene",
    "customer_type",
    "title",
    "content",
    "tags",
    "is_active",
    "审核状态",
    "审核人",
    "备注",
]


@dataclass(frozen=True)
class SeedScriptTemplate:
    sequence: int
    scene: str
    customer_type: str | None
    title: str
    content: str
    tags: list[str]
    is_active: bool
    review_status: str
    reviewer: str | None
    note: str

    @property
    def is_approved(self) -> bool:
        return self.review_status == "approved"


@dataclass(frozen=True)
class ImportResult:
    total: int
    created: int
    updated: int


def load_templates_from_markdown(path: Path = DEFAULT_DATASET) -> list[SeedScriptTemplate]:
    rows = _read_markdown_table(path)
    templates: list[SeedScriptTemplate] = []
    for row in rows:
        sequence = int(row["序号"])
        scene = row["scene"].strip()
        title = row["title"].strip()
        content = row["content"].strip()
        if not scene:
            raise ValueError(f"Missing scene at row {sequence}")
        if not title:
            raise ValueError(f"Missing title at row {sequence}")
        if not content:
            raise ValueError(f"Missing content at row {sequence}")
        templates.append(
            SeedScriptTemplate(
                sequence=sequence,
                scene=scene,
                customer_type=row["customer_type"].strip() or None,
                title=title,
                content=content,
                tags=_parse_tags(row["tags"]),
                is_active=_parse_bool(row["is_active"]),
                review_status=_normalize_review_status(row["审核状态"], row["备注"]),
                reviewer=row["审核人"].strip() or None,
                note=row["备注"].strip(),
            )
        )
    return templates


def import_templates_from_markdown(
    path: Path = DEFAULT_DATASET,
    *,
    actor_id: str = ACTOR_ID,
) -> ImportResult:
    templates = load_templates_from_markdown(path)
    compliance_service = ComplianceCheckService()
    created = 0
    updated = 0

    with AgentSessionLocal() as session:
        for seed_template in templates:
            check_result = compliance_service.check_text(
                ComplianceCheckRequest(text=seed_template.content, scene=seed_template.scene)
            )
            if check_result.risk_level == "BLOCK":
                raise ValueError(f"Template row {seed_template.sequence} failed BLOCK compliance check")

            existing = (
                session.query(ScriptTemplate)
                .filter(
                    ScriptTemplate.created_by == actor_id,
                    ScriptTemplate.scene == seed_template.scene,
                    ScriptTemplate.title == seed_template.title,
                )
                .one_or_none()
            )
            values = {
                "customer_type": seed_template.customer_type,
                "content": seed_template.content,
                "tags": seed_template.tags,
                "is_active": seed_template.is_active,
                "is_approved": seed_template.is_approved,
                "approved_by": seed_template.reviewer if seed_template.is_approved else None,
                "approved_at": datetime.now(timezone.utc).replace(tzinfo=None) if seed_template.is_approved else None,
                "updated_by": actor_id,
            }
            if existing is None:
                session.add(
                    ScriptTemplate(
                        scene=seed_template.scene,
                        title=seed_template.title,
                        created_by=actor_id,
                        version=1,
                        usage_count=0,
                        **values,
                    )
                )
                created += 1
            else:
                if _content_changed(existing, values):
                    existing.version += 1
                for key, value in values.items():
                    setattr(existing, key, value)
                updated += 1
        session.commit()

    return ImportResult(total=len(templates), created=created, updated=updated)


def _content_changed(existing: ScriptTemplate, values: dict) -> bool:
    return any(
        [
            existing.customer_type != values["customer_type"],
            existing.content != values["content"],
            existing.tags != values["tags"],
            existing.is_active != values["is_active"],
            existing.is_approved != values["is_approved"],
        ]
    )


def _read_markdown_table(path: Path) -> list[dict[str, str]]:
    if not path.exists():
        raise FileNotFoundError(path)

    table_lines = [
        line.strip()
        for line in path.read_text(encoding="utf-8").splitlines()
        if line.strip().startswith("|") and line.strip().endswith("|")
    ]
    if len(table_lines) < 3:
        raise ValueError("Markdown table is missing or empty")

    headers = _split_markdown_row(table_lines[0])
    if headers != REQUIRED_HEADERS:
        raise ValueError(f"Unexpected table headers: {headers}")

    rows: list[dict[str, str]] = []
    for line in table_lines[2:]:
        cells = _split_markdown_row(line)
        if len(cells) != len(headers):
            raise ValueError(f"Column count mismatch: {line}")
        rows.append(dict(zip(headers, cells, strict=True)))
    return rows


def _split_markdown_row(line: str) -> list[str]:
    return [cell.strip() for cell in line.strip().strip("|").split("|")]


def _parse_tags(value: str) -> list[str]:
    return [tag.strip() for tag in value.split(";") if tag.strip()]


def _parse_bool(value: str) -> bool:
    normalized = value.strip().lower()
    if normalized == "true":
        return True
    if normalized == "false":
        return False
    raise ValueError(f"Expected true/false, got: {value}")


def _normalize_review_status(value: str, note: str) -> str:
    if "已审核" in value:
        return "approved"
    if "测试数据" in value or "测试数据" in note:
        return "test_data"
    return "pending"


def main() -> None:
    parser = argparse.ArgumentParser(description="Import Sprint 2 script templates.")
    parser.add_argument("--dataset", type=Path, default=DEFAULT_DATASET)
    args = parser.parse_args()
    result = import_templates_from_markdown(args.dataset)
    print(f"Imported script templates: total={result.total}, created={result.created}, updated={result.updated}")


if __name__ == "__main__":
    main()