monitor2.py

import numpy as np
import pandas as pd


def format_trade_monitor_result(
    result,
    *,
    country_col,
    value_label="取引額",
    quantity_label="統計数量",
    value_unit=None,
    quantity_unit=None,
    pct_digits=1,
    number_digits=0,
):
    """
    monitor_trade() の返り値を、人間が読みやすい日本語表記に変換する。

    元の result は変更しない。

    Parameters
    ----------
    result : dict
        monitor_trade() の返り値

    country_col : str
        原産国カラム名
        例: "原産国"

    value_label : str
        金額の表示名
        例: "取引額", "課税価格"

    quantity_label : str
        数量の表示名
        例: "統計数量"

    value_unit : str or None
        金額単位
        例: "円", "千円"

    quantity_unit : str or None
        数量単位
        例: "KG", "MT", "NO"

    pct_digits : int
        %表示の小数点桁数

    number_digits : int
        金額・数量表示の小数点桁数

    Returns
    -------
    dict

    {
        "概要": DataFrame,
        "月次推移": DataFrame,
        "全体指標": DataFrame,
        "集中度": DataFrame,
        "国別一覧": DataFrame,
        "金額TOP": DataFrame,
        "数量TOP": DataFrame,
        "新規・消失国": DataFrame,
    }
    """

    # ============================================================
    # 表示用ヘルパー
    # ============================================================

    def is_missing(x):
        try:
            return pd.isna(x)
        except Exception:
            return False

    def fmt_number(x, unit=None, signed=False):
        if is_missing(x):
            return "-"

        if np.isposinf(x):
            return "+∞"

        if np.isneginf(x):
            return "-∞"

        sign = "+" if signed and x > 0 else ""

        text = f"{sign}{x:,.{number_digits}f}"

        if unit:
            text += f" {unit}"

        return text

    def fmt_pct(x):
        """
        0.253 -> +25.3%
        -0.12 -> -12.0%
        """
        if is_missing(x):
            return "-"

        if np.isposinf(x):
            return "+∞"

        if np.isneginf(x):
            return "-∞"

        return f"{x * 100:+.{pct_digits}f}%"

    def fmt_share(x):
        """
        シェア用。
        0.253 -> 25.3%
        """
        if is_missing(x):
            return "-"

        return f"{x * 100:.{pct_digits}f}%"

    def fmt_pp(x):
        """
        既に percentage point に変換済みの値。
        15.2 -> +15.2pt
        """
        if is_missing(x):
            return "-"

        return f"{x:+.{pct_digits}f}pt"

    def fmt_z(x):
        if is_missing(x):
            return "-"

        if np.isposinf(x):
            return "+∞"

        if np.isneginf(x):
            return "-∞"

        return f"{x:+.2f}"

    def fmt_hhi(x):
        if is_missing(x):
            return "-"

        return f"{x:.3f}"

    def fmt_effective_supplier(x):
        if is_missing(x):
            return "-"

        return f"{x:.2f}"

    def fmt_count(x):
        if is_missing(x):
            return "-"

        return f"{int(x):,}"

    def fmt_rank(x):
        if is_missing(x):
            return "-"

        return f"{int(x)}位"

    def fmt_rank_change(x):
        if is_missing(x):
            return "-"

        x = int(x)

        if x > 0:
            return f"{x}位上昇"

        if x < 0:
            return f"{abs(x)}位低下"

        return "変化なし"

    def fmt_status(x):
        mapping = {
            "new": "新規",
            "disappeared": "消失",
            "continuing": "継続",
        }

        return mapping.get(x, x)

    # ============================================================
    # 0. 概要
    # ============================================================

    warnings = result.get("warnings", [])

    summary = pd.DataFrame(
        {
            "対象月": [
                result.get("target_month", "-")
            ],
            "比較対象（前年同月）": [
                result.get("yoy_month", "-")
            ],
            "警告": [
                " / ".join(warnings)
                if warnings
                else "なし"
            ],
        }
    )

    # ============================================================
    # 1. 月次推移
    # ============================================================

    monthly = result["monthly"].copy()

    monthly_columns = {
        "month": "年月",
        "value": value_label,
        "quantity": quantity_label,
        "unit_price": "平均単価",
    }

    monthly = monthly.rename(
        columns=monthly_columns
    )

    if value_label in monthly.columns:
        monthly[value_label] = monthly[
            value_label
        ].map(
            lambda x: fmt_number(
                x,
                value_unit
            )
        )

    if quantity_label in monthly.columns:
        monthly[quantity_label] = monthly[
            quantity_label
        ].map(
            lambda x: fmt_number(
                x,
                quantity_unit
            )
        )

    unit_price_unit = None

    if value_unit and quantity_unit:
        unit_price_unit = (
            f"{value_unit}/{quantity_unit}"
        )

    if "平均単価" in monthly.columns:
        monthly["平均単価"] = monthly[
            "平均単価"
        ].map(
            lambda x: fmt_number(
                x,
                unit_price_unit
            )
        )

    monthly["年月"] = (
        monthly["年月"]
        .astype(str)
    )

    # ============================================================
    # 2. 全体指標
    # ============================================================

    metric = result[
        "metric_summary"
    ].copy()

    metric_names = {
        "value": value_label,
        "quantity": quantity_label,
        "unit_price": "平均単価",
    }

    metric_rows = []

    for metric_name, display_name in metric_names.items():

        if metric_name not in metric.index:
            continue

        row = metric.loc[
            metric_name
        ]

        if metric_name == "value":
            unit = value_unit

        elif metric_name == "quantity":
            unit = quantity_unit

        else:
            unit = unit_price_unit

        metric_rows.append(
            {
                "指標":
                    display_name,

                "当月値":
                    fmt_number(
                        row["current"],
                        unit
                    ),

                "前年同月値":
                    fmt_number(
                        row["yoy_base"],
                        unit
                    ),

                "前年同月比":
                    fmt_pct(
                        row["yoy_pct"]
                    ),

                "前月比":
                    fmt_pct(
                        row["mom_pct"]
                    ),

                "直近3か月平均比":
                    fmt_pct(
                        row[
                            "vs_prev3m_avg_pct"
                        ]
                    ),

                "過去12か月中央値比":
                    fmt_pct(
                        row[
                            "vs_prev12m_median_pct"
                        ]
                    ),

                "異常度（Robust Z）":
                    fmt_z(
                        row[
                            "robust_z_12m"
                        ]
                    ),
            }
        )

    metric_readable = pd.DataFrame(
        metric_rows
    )

    # ============================================================
    # 3. 集中度
    # ============================================================

    concentration = result[
        "concentration"
    ].copy()

    concentration_rows = []

    basis_names = {
        "value": "金額ベース",
        "quantity": "数量ベース",
    }

    for basis, display_name in basis_names.items():

        if basis not in concentration.index:
            continue

        row = concentration.loc[
            basis
        ]

        concentration_rows.append(
            {
                "集計基準":
                    display_name,

                "CR1（当月）":
                    fmt_share(
                        row["cr1_current"]
                    ),

                "CR1（前年同月）":
                    fmt_share(
                        row["cr1_yoy_base"]
                    ),

                "CR1前年差":
                    fmt_pp(
                        row["cr1_delta_pp"]
                    ),

                "CR3（当月）":
                    fmt_share(
                        row["cr3_current"]
                    ),

                "CR3（前年同月）":
                    fmt_share(
                        row["cr3_yoy_base"]
                    ),

                "CR3前年差":
                    fmt_pp(
                        row["cr3_delta_pp"]
                    ),

                "CR5（当月）":
                    fmt_share(
                        row["cr5_current"]
                    ),

                "CR5（前年同月）":
                    fmt_share(
                        row["cr5_yoy_base"]
                    ),

                "CR5前年差":
                    fmt_pp(
                        row["cr5_delta_pp"]
                    ),

                "HHI（当月）":
                    fmt_hhi(
                        row["hhi_current"]
                    ),

                "HHI（前年同月）":
                    fmt_hhi(
                        row["hhi_yoy_base"]
                    ),

                "HHI前年差":
                    fmt_hhi(
                        row["hhi_delta"]
                    ),

                "有効供給国数（当月）":
                    fmt_effective_supplier(
                        row[
                            "effective_suppliers_current"
                        ]
                    ),

                "有効供給国数（前年同月）":
                    fmt_effective_supplier(
                        row[
                            "effective_suppliers_yoy_base"
                        ]
                    ),

                "実供給国数（当月）":
                    fmt_count(
                        row[
                            "supplier_count_current"
                        ]
                    ),

                "実供給国数（前年同月）":
                    fmt_count(
                        row[
                            "supplier_count_yoy_base"
                        ]
                    ),

                "供給構造変化指数":
                    fmt_share(
                        row[
                            "structure_change_index"
                        ]
                    ),
            }
        )

    concentration_readable = pd.DataFrame(
        concentration_rows
    )

    # ============================================================
    # 4. 国別テーブル共通処理
    # ============================================================

    def format_country_table(df):

        df = df.copy()

        # -------------------------------
        # 数値フォーマット
        # -------------------------------

        value_number_cols = [
            "value_current",
            "value_yoy_base",
        ]

        value_signed_cols = [
            "value_delta",
        ]

        quantity_number_cols = [
            "quantity_current",
            "quantity_yoy_base",
        ]

        quantity_signed_cols = [
            "quantity_delta",
        ]

        pct_cols = [
            "value_yoy_pct",
            "quantity_yoy_pct",
            "unit_price_yoy_pct",
        ]

        share_cols = [
            "value_share_current",
            "value_share_yoy_base",
            "quantity_share_current",
            "quantity_share_yoy_base",
        ]

        pp_cols = [
            "value_share_delta_pp",
            "quantity_share_delta_pp",
        ]

        rank_cols = [
            "value_rank_current",
            "value_rank_yoy_base",
            "quantity_rank_current",
            "quantity_rank_yoy_base",
        ]

        rank_change_cols = [
            "value_rank_change",
            "quantity_rank_change",
        ]

        unit_price_cols = [
            "unit_price_current",
            "unit_price_yoy_base",
        ]

        for col in value_number_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    lambda x: fmt_number(
                        x,
                        value_unit
                    )
                )

        for col in value_signed_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    lambda x: fmt_number(
                        x,
                        value_unit,
                        signed=True,
                    )
                )

        for col in quantity_number_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    lambda x: fmt_number(
                        x,
                        quantity_unit
                    )
                )

        for col in quantity_signed_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    lambda x: fmt_number(
                        x,
                        quantity_unit,
                        signed=True,
                    )
                )

        for col in pct_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    fmt_pct
                )

        for col in share_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    fmt_share
                )

        for col in pp_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    fmt_pp
                )

        for col in rank_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    fmt_rank
                )

        for col in rank_change_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    fmt_rank_change
                )

        for col in unit_price_cols:
            if col in df.columns:
                df[col] = df[col].map(
                    lambda x: fmt_number(
                        x,
                        unit_price_unit
                    )
                )

        if "status" in df.columns:
            df["status"] = df[
                "status"
            ].map(
                fmt_status
            )

        if "event_flag" in df.columns:
            df["event_flag"] = np.where(
                df["event_flag"],
                "注目",
                "",
            )

        # -------------------------------
        # 日本語ヘッダー
        # -------------------------------

        column_names = {

            country_col:
                country_col,

            "value_current":
                f"{value_label}（当月）",

            "value_yoy_base":
                f"{value_label}（前年同月）",

            "value_delta":
                f"{value_label}前年差",

            "value_yoy_pct":
                f"{value_label}前年同月比",

            "value_share_current":
                f"{value_label}シェア（当月）",

            "value_share_yoy_base":
                f"{value_label}シェア（前年同月）",

            "value_share_delta_pp":
                f"{value_label}シェア前年差",

            "value_rank_current":
                "金額順位（当月）",

            "value_rank_yoy_base":
                "金額順位（前年同月）",

            "value_rank_change":
                "金額順位変化",

            "quantity_current":
                f"{quantity_label}（当月）",

            "quantity_yoy_base":
                f"{quantity_label}（前年同月）",

            "quantity_delta":
                f"{quantity_label}前年差",

            "quantity_yoy_pct":
                f"{quantity_label}前年同月比",

            "quantity_share_current":
                f"{quantity_label}シェア（当月）",

            "quantity_share_yoy_base":
                f"{quantity_label}シェア（前年同月）",

            "quantity_share_delta_pp":
                f"{quantity_label}シェア前年差",

            "quantity_rank_current":
                "数量順位（当月）",

            "quantity_rank_yoy_base":
                "数量順位（前年同月）",

            "quantity_rank_change":
                "数量順位変化",

            "unit_price_current":
                "平均単価（当月）",

            "unit_price_yoy_base":
                "平均単価（前年同月）",

            "unit_price_yoy_pct":
                "平均単価前年同月比",

            "status":
                "供給状態",

            "event_flag":
                "注目フラグ",
        }

        df = df.rename(
            columns=column_names
        )

        # -------------------------------
        # 列順を人間向けに整理
        # -------------------------------

        desired_order = [

            country_col,

            f"{value_label}（当月）",
            f"{value_label}（前年同月）",
            f"{value_label}前年差",
            f"{value_label}前年同月比",

            f"{value_label}シェア（当月）",
            f"{value_label}シェア（前年同月）",
            f"{value_label}シェア前年差",

            "金額順位（当月）",
            "金額順位（前年同月）",
            "金額順位変化",

            f"{quantity_label}（当月）",
            f"{quantity_label}（前年同月）",
            f"{quantity_label}前年差",
            f"{quantity_label}前年同月比",

            f"{quantity_label}シェア（当月）",
            f"{quantity_label}シェア（前年同月）",
            f"{quantity_label}シェア前年差",

            "数量順位（当月）",
            "数量順位（前年同月）",
            "数量順位変化",

            "平均単価（当月）",
            "平均単価（前年同月）",
            "平均単価前年同月比",

            "供給状態",
            "注目フラグ",
        ]

        desired_order = [
            col
            for col in desired_order
            if col in df.columns
        ]

        other_cols = [
            col
            for col in df.columns
            if col not in desired_order
        ]

        return df[
            desired_order
            + other_cols
        ].reset_index(drop=True)

    # ============================================================
    # 5. 各国別テーブル
    # ============================================================

    country_all = format_country_table(
        result["country_all"]
    )

    country_top_value = format_country_table(
        result["country_top_value"]
    )

    country_top_quantity = format_country_table(
        result["country_top_quantity"]
    )

    events = format_country_table(
        result["events"]
    )

    # ============================================================
    # Return
    # ============================================================

    return {
        "概要":
            summary,

        "月次推移":
            monthly,

        "全体指標":
            metric_readable,

        "集中度":
            concentration_readable,

        "国別一覧":
            country_all,

        "金額TOP":
            country_top_value,

        "数量TOP":
            country_top_quantity,

        "新規・消失国":
            events,
    }