app.py

import requests
import pandas as pd
import streamlit as st
import plotly.express as px
import plotly.graph_objects as go
import matplotlib.font_manager as fm
import os
import platform

# ── 頁面設定 ──────────────────────────────────────────────
st.set_page_config(
    page_title="MOMO 商品價格分析",
    page_icon="🛒",
    layout="wide",
)

st.title("🛒 MOMO 電商商品價格分析儀表板")
st.markdown("輸入關鍵字，即時抓取 MOMO 商品資料並視覺化分析。")

# ── 中文字型安裝（不使用 wget） ────────────────────────────
@st.cache_resource
def install_chinese_font():
    """
    嘗試安裝系統中文字型；若找不到則下載 Noto Sans TC。
    回傳字型路徑（供 matplotlib 使用，Plotly 另行設定）。
    """
    # 常見系統中文字型路徑
    system_fonts = {
        "Linux": [
            "/usr/share/fonts/truetype/noto/NotoSansCJK-Regular.ttc",
            "/usr/share/fonts/opentype/noto/NotoSansCJKtc-Regular.otf",
        ],
        "Darwin": [  # macOS
            "/System/Library/Fonts/PingFang.ttc",
            "/Library/Fonts/Arial Unicode.ttf",
        ],
        "Windows": [
            "C:/Windows/Fonts/msjh.ttc",
            "C:/Windows/Fonts/mingliu.ttc",
        ],
    }
    sys_name = platform.system()
    for path in system_fonts.get(sys_name, []):
        if os.path.exists(path):
            return path

    # 找不到系統字型 → 用 requests 下載 Noto Sans TC
    font_path = "NotoSansTC-Regular.ttf"
    if not os.path.exists(font_path):
        url = (
            "https://github.com/googlefonts/noto-cjk/raw/main/Sans/OTF/TraditionalChinese/"
            "NotoSansCJKtc-Regular.otf"
        )
        try:
            r = requests.get(url, timeout=30)
            r.raise_for_status()
            with open(font_path, "wb") as f:
                f.write(r.content)
            fm.fontManager.addfont(font_path)
        except Exception as e:
            st.warning(f"字型下載失敗：{e}，圖表中文可能顯示異常。")
            return None
    return font_path


install_chinese_font()


# ── 抓取 MOMO 資料 ─────────────────────────────────────────
def fetch_momo(keyword: str, max_items: int) -> pd.DataFrame:
    url = "https://apisearch.momoshop.com.tw/momoSearchCloud/moec/textSearch"
    headers = {
        "User-Agent": (
            "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
            "AppleWebKit/537.36 (KHTML, like Gecko) "
            "Chrome/127.0.0.0 Safari/537.36"
        )
    }

    all_products = []
    page = 1
    per_page = 24  # MOMO 預設每頁約 24 筆

    while len(all_products) < max_items:
        payload = {
            "host": "momoshop",
            "flag": "searchEngine",
            "data": {
                "specialGoodsType": "",
                "isBrandSeriesPage": "false",
                "authorNo": "",
                "originalCateCode": "",
                "cateType": "",
                "searchValue": keyword,
                "cateCode": "",
                "cateLevel": "-1",
                "cp": "N", "NAM": "N", "first": "N", "freeze": "N",
                "superstore": "N", "tvshop": "N", "china": "N",
                "tomorrow": "N", "stockYN": "N", "prefere": "N",
                "threeHours": "N", "video": "N", "cycle": "N",
                "cod": "N", "superstorePay": "N",
                "showType": "chessboardType",
                "curPage": str(page),
                "priceS": "0",
                "priceE": "9999999",
                "searchType": "1",
                "reduceKeyword": "",
                "isFuzzy": "0",
                "rtnCateDatainfo": {
                    "cateCode": "",
                    "cateLv": "-1",
                    "keyword": keyword,
                    "curPage": str(page),
                    "historyDoPush": "false",
                    "timestamp": 1723036027826,
                },
                "flag": 2018,
                "serviceCode": "MT01",
                "addressSearchData": {},
                "adSource": "tenmax",
            },
        }

        try:
            resp = requests.post(url, headers=headers, json=payload, timeout=15)
            resp.raise_for_status()
        except Exception as e:
            st.error(f"第 {page} 頁請求失敗：{e}")
            break

        data = resp.json()
        products = data.get("rtnSearchData", {}).get("goodsInfoList", [])
        if not products:
            break

        for p in products:
            name = p.get("goodsName", "")
            price_raw = p.get("goodsPrice", "0")
            try:
                price = int(
                    str(price_raw).replace("$", "").replace(",", "").strip()
                )
            except ValueError:
                price = 0
            all_products.append({"品名": name, "價格": price})

        page += 1
        if len(products) < per_page:
            break  # 已無更多頁

    df = pd.DataFrame(all_products[:max_items])
    return df


# ── 側邊欄：使用者輸入 ────────────────────────────────────
with st.sidebar:
    st.header("⚙️ 查詢設定")
    keyword = st.text_input("搜尋關鍵字", value="耳機")
    max_items = st.slider("最多筆數", min_value=10, max_value=200, value=70, step=10)
    search_btn = st.button("🔍 開始抓取", use_container_width=True)
    st.divider()
    st.markdown("資料來源：[MOMO 購物網](https://www.momoshop.com.tw/)")

# ── 主流程 ────────────────────────────────────────────────
if search_btn:
    if not keyword.strip():
        st.warning("請輸入搜尋關鍵字！")
        st.stop()

    with st.spinner(f"正在抓取「{keyword}」的商品資料，請稍候..."):
        df = fetch_momo(keyword, max_items)

    if df.empty:
        st.error("未抓到任何資料，請檢查網路或更換關鍵字。")
        st.stop()

    # 統計資訊
    avg_price = df["價格"].mean()
    max_price = df["價格"].max()
    min_price = df["價格"].min()

    st.success(f"共抓取 **{len(df)}** 筆資料")

    col1, col2, col3 = st.columns(3)
    col1.metric("💰 平均價格", f"${avg_price:,.0f}")
    col2.metric("📈 最高價格", f"${max_price:,.0f}")
    col3.metric("📉 最低價格", f"${min_price:,.0f}")

    st.divider()

    # ── 圖表區 ────────────────────────────────────────────
    tab1, tab2, tab3 = st.tabs(["📈 折線圖", "🥧 圓餅圖", "☀️ 旭日圖"])

    # ---- Tab1：折線圖 ----
    with tab1:
        st.subheader(f"「{keyword}」商品售價折線圖")
        df_line = df.copy().reset_index()
        df_line.columns = ["商品序號", "品名", "價格"]

        fig_line = go.Figure()
        fig_line.add_trace(
            go.Scatter(
                x=df_line["商品序號"],
                y=df_line["價格"],
                mode="lines+markers",
                name="售價",
                line=dict(color="#3B82F6", width=2),
                marker=dict(size=5),
                hovertext=df_line["品名"],
                hovertemplate="<b>%{hovertext}</b><br>價格：$%{y:,}<extra></extra>",
            )
        )
        fig_line.add_hline(
            y=avg_price,
            line_dash="dash",
            line_color="red",
            annotation_text=f"平均 ${avg_price:,.0f}",
            annotation_position="bottom right",
        )
        fig_line.update_layout(
            xaxis_title="商品序號",
            yaxis_title="價格 (NTD)",
            height=500,
            hovermode="x unified",
            plot_bgcolor="rgba(0,0,0,0)",
            paper_bgcolor="rgba(0,0,0,0)",
        )
        st.plotly_chart(fig_line, use_container_width=True)

    # ---- Tab2：圓餅圖（價格區間分布） ----
    with tab2:
        st.subheader(f"「{keyword}」價格區間分布圓餅圖")

        bins = [0, 500, 1000, 3000, 5000, 10000, float("inf")]
        labels = ["≤500", "501~1000", "1001~3000", "3001~5000", "5001~10000", ">10000"]
        df["價格區間"] = pd.cut(df["價格"], bins=bins, labels=labels, right=True)
        pie_data = df["價格區間"].value_counts().reset_index()
        pie_data.columns = ["價格區間", "數量"]

        fig_pie = px.pie(
            pie_data,
            names="價格區間",
            values="數量",
            title=f"{keyword} 各價格區間商品數量",
            hole=0.35,
            color_discrete_sequence=px.colors.qualitative.Pastel,
        )
        fig_pie.update_traces(textposition="inside", textinfo="percent+label")
        fig_pie.update_layout(
            height=500,
            plot_bgcolor="rgba(0,0,0,0)",
            paper_bgcolor="rgba(0,0,0,0)",
        )
        st.plotly_chart(fig_pie, use_container_width=True)

    # ---- Tab3：旭日圖（品牌 × 價格區間） ----
    with tab3:
        st.subheader(f"「{keyword}」品牌 × 價格區間旭日圖")
        st.caption("品牌由品名前幾個字推斷，僅供參考。")

        # 從品名前 N 字切出「品牌」（簡易規則）
        def extract_brand(name: str, n: int = 4) -> str:
            return name[:n] if len(name) >= n else name

        df["品牌"] = df["品名"].apply(extract_brand)
        sun_data = (
            df.groupby(["價格區間", "品牌"])
            .size()
            .reset_index(name="數量")
        )

        fig_sun = px.sunburst(
            sun_data,
            path=["價格區間", "品牌"],
            values="數量",
            title=f"{keyword} 價格區間 × 品牌旭日圖",
            color="數量",
            color_continuous_scale="RdBu",
        )
        fig_sun.update_layout(
            height=600,
            plot_bgcolor="rgba(0,0,0,0)",
            paper_bgcolor="rgba(0,0,0,0)",
        )
        st.plotly_chart(fig_sun, use_container_width=True)

    st.divider()

    # ── 原始資料表 ─────────────────────────────────────────
    with st.expander("📋 查看原始資料", expanded=False):
        st.dataframe(df[["品名", "價格"]].reset_index(drop=True), use_container_width=True)

    # ── 下載 CSV ──────────────────────────────────────────
    csv = df[["品名", "價格"]].to_csv(index=False, encoding="utf-8-sig")
    st.download_button(
        label="⬇️ 下載 CSV",
        data=csv,
        file_name=f"MOMO_{keyword}.csv",
        mime="text/csv",
    )