from __future__ import annotations

"""
Campaign Name 12 字段解析器。

按 '_' 分割 campaign_name，提取 12 个标准字段，
自动归类市场/funnel/平台/优化目标/产品线/受众类型。
字段内部用 '-' 连接（如 Bone-Conduction），不与 '_' 分隔符混淆。

位置:  1      2              3      4      5          6               7        8           9       10        11        12
字段:  市场   funnel          平台    渠道   优化目标    产品线          产品型号   项目类型     受众分类  素材类型   日期       备注
示例:  US    Conversion     Meta    All    Purchase   Bone-Conduction  Others   AO          ASC     Mix       20260123  coreNumber
"""
import logging
from typing import Any

import pandas as pd

log = logging.getLogger("ad-agent")

VALID_MARKETS = {"US", "CA"}
VALID_FUNNELS = {"Awareness", "Consideration", "Conversion"}
VALID_PLATFORMS = {"Meta", "Google", "Criteo", "Bing", "TikTok"}
VALID_OBJECTIVES = {
    "Purchase", "Lead", "ATC", "AtoB",
    "Linkclick", "Buttonclick", "Pageview", "PE",
}

AUDIENCE_REMARKETING = "remarketing"
AUDIENCE_ASC = "asc"
AUDIENCE_COLD = "cold"

_FIELD_KEYS = [
    "market", "funnel", "parsed_platform", "channel",
    "ad_objective", "product_line", "product_model", "project_type",
    "audience_raw", "creative_type", "campaign_date", "note",
]

_OUTPUT_COLUMNS = _FIELD_KEYS + ["audience_type", "parse_ok", "objective_group"]


def _classify_audience(raw: str) -> str:
    upper = raw.upper()
    if upper == "RMKT":
        return AUDIENCE_REMARKETING
    if upper == "ASC":
        return AUDIENCE_ASC
    return AUDIENCE_COLD


def _derive_objective_group(objective: str | None, audience: str) -> str:
    if objective == "Purchase":
        if audience == AUDIENCE_REMARKETING:
            return "Purchase_RMKT"
        if audience == AUDIENCE_ASC:
            return "Purchase_ASC"
        return "Purchase_Cold"
    if objective in VALID_OBJECTIVES:
        return objective
    return "Other"


def parse_campaign_name(name: str) -> dict[str, Any]:
    """
    解析单个 campaign name，返回 12 个原始字段 + 派生字段。
    不足 12 段时缺失字段为 None；超出 12 段时多余部分合并到 note。
    """
    parts = str(name).split("_")

    result: dict[str, Any] = {}
    for i, key in enumerate(_FIELD_KEYS):
        if i < len(parts):
            if key == "note" and len(parts) > 12:
                result[key] = "_".join(parts[11:])
            else:
                result[key] = parts[i]
        else:
            result[key] = None

    result["audience_type"] = (
        _classify_audience(result["audience_raw"])
        if result["audience_raw"]
        else AUDIENCE_COLD
    )

    result["parse_ok"] = (
        len(parts) >= 5
        and result.get("market") in VALID_MARKETS
        and result.get("funnel") in VALID_FUNNELS
    )

    result["objective_group"] = _derive_objective_group(
        result.get("ad_objective"), result["audience_type"]
    )

    return result


def enrich_dataframe(df: pd.DataFrame) -> pd.DataFrame:
    """
    批量解析 campaign_name 列，新增解析字段列。返回新 DataFrame。
    """
    if "campaign_name" not in df.columns:
        log.warning("DataFrame 中没有 campaign_name 列，跳过解析")
        return df

    df = df.copy()
    parsed = df["campaign_name"].apply(parse_campaign_name)
    parsed_df = pd.DataFrame(parsed.tolist(), index=df.index)

    for col in _OUTPUT_COLUMNS:
        if col in parsed_df.columns:
            df[col] = parsed_df[col]

    ok_count = df["parse_ok"].sum() if "parse_ok" in df.columns else 0
    total = len(df)
    if total > 0:
        rate = ok_count / total * 100
        log.info(f"Campaign name 解析: {ok_count}/{total} 成功 ({rate:.1f}%)")
        if rate < 90:
            log.warning("解析成功率低于 90%，可能存在命名不规范的 campaign")

    return df