后台店铺数据重复检查独立成菜单:删除店铺导入测试、新增撞款详情卡片区

- 删除「店铺导入测试」菜单及全部实现:admin.html 面板、admin.js 逻辑、
  admin_api.py 路由/常量/权限映射、Java 默认菜单
- 「店铺数据记录」页内子 tab 移除,只保留店铺数据列表
- 「店铺管理」下新增「店铺数据重复检查」菜单(V106 导入测试菜单保留并新增
  V107 清理 + 幂等插入新菜单,Java 清单同步)
- 新面板:筛选/指标卡/店铺上架分布/矩阵表格/撞款详情卡片区(同 ASIN 跨店
  上架明细,查看明细开抽屉,上架时间倒序)
- 重复扫描结果改为 MySQL 缓存 + 每日定时扫描(start_duplicate_scan_scheduler)
- 新增 test_admin_shop_data_duplicate_check.py 单元测试
This commit is contained in:
2026-09-04 00:16:14 +08:00
parent 672b9f9b46
commit d0db577a9c
9 changed files with 2693 additions and 514 deletions
@@ -71,6 +71,7 @@ public class PermissionMenuSchemaInitializer {
new DefaultAdminMenu("店铺管理", "admin_shop_manage", "shop-manage", 50, "admin_group_shop"),
new DefaultAdminMenu("最低价ASIN设置", "admin_skip_price_asin", "skip-price-asin", 60, "admin_group_shop"),
new DefaultAdminMenu("店铺数据记录", "admin_shop_data_crawl_tasks", "shop-data-crawl-tasks", 82, "admin_group_shop"),
new DefaultAdminMenu("店铺数据重复检查", "admin_shop_data_duplicate_check", "shop-data-duplicate-check", 84, "admin_group_shop"),
new DefaultAdminMenu("生成记录", "admin_history", "history", 70, "admin_group_record"),
new DefaultAdminMenu("视频任务记录", "admin_image_video_tasks", "image-video-tasks", 75, "admin_group_record"),
new DefaultAdminMenu("软件版本管理", "admin_version", "version", 80, "admin_group_record"),
@@ -0,0 +1,9 @@
-- V106: 新增「店铺导入测试」后台菜单(用于重复 ASIN 结果模拟验证)
-- 幂等:仅当 column_key 不存在时插入,绑定到「店铺管理」分组(admin_group_shop)。
INSERT INTO columns (name, column_key, menu_type, route_path, sort_order, parent_id)
SELECT '店铺导入测试', 'admin_shop_data_import', 'admin', 'shop-data-import', 83, parent.id
FROM columns parent
WHERE parent.column_key = 'admin_group_shop'
AND NOT EXISTS (
SELECT 1 FROM columns WHERE column_key = 'admin_shop_data_import'
);
@@ -0,0 +1,18 @@
-- V107: 「店铺导入测试」菜单下线,新增「店铺数据重复检查」菜单
-- 1) 删除「店铺导入测试」(admin_shop_data_import)后台菜单行与其授权记录(幂等)
DELETE FROM user_column_permission
WHERE column_id IN (
SELECT id FROM columns WHERE column_key = 'admin_shop_data_import'
);
DELETE FROM columns WHERE column_key = 'admin_shop_data_import';
-- 2) 新增「店铺数据重复检查」菜单,挂到「店铺管理」分组(admin_group_shop
-- 幂等:仅当 column_key 不存在时插入。
INSERT INTO columns (name, column_key, menu_type, route_path, sort_order, parent_id)
SELECT '店铺数据重复检查', 'admin_shop_data_duplicate_check', 'admin', 'shop-data-duplicate-check', 84, parent.id
FROM columns parent
WHERE parent.column_key = 'admin_group_shop'
AND NOT EXISTS (
SELECT 1 FROM columns WHERE column_key = 'admin_shop_data_duplicate_check'
);
+3 -1
View File
@@ -12,7 +12,7 @@ from flask_cors import CORS
from utils.db import init_db
from blueprints.auth import auth
from blueprints.main import main
from blueprints.admin_api import admin_api
from blueprints.admin_api import admin_api, start_duplicate_scan_scheduler
from blueprints.version import version_bp
BASE_DIR = os.path.dirname(os.path.abspath(__file__))
@@ -50,6 +50,8 @@ app.register_blueprint(version_bp)
def run_app(host='0.0.0.0', port=15124):
init_db()
# 每日凌晨全量扫描重复 ASIN,结果缓存 MySQL,页面读取缓存不再实时拉取 Excel
start_duplicate_scan_scheduler(app)
app.run(host=host, port=port, threaded=True, use_reloader=False)
+722 -75
View File
@@ -8,8 +8,9 @@ import re
import secrets
import tempfile
import threading
import time
import zipfile
from datetime import datetime
from datetime import datetime, timedelta
from pathlib import Path
from urllib.parse import quote
@@ -52,6 +53,49 @@ IMAGE_VIDEO_DATA_PERMISSION_KEY = 'admin_image_video_task_data'
SHOP_DATA_CRAWL_DATA_PERMISSION_KEY = 'admin_shop_data_crawl_task_data'
# ---------- 拼音首字母(分组管理通讯录用,无第三方依赖) ----------
# GB2312 区位码按拼音音序排列,用码位区间即可精确判定首字母;仅覆盖 GB2312 汉字。
_PYINYIN_BOUNDARIES = [
('A', 0xB0A1), ('B', 0xB0C5), ('C', 0xB2C1), ('D', 0xB4EE), ('E', 0xB6EA),
('F', 0xB7A2), ('G', 0xB8C1), ('H', 0xB9FE), ('J', 0xBBF7), ('K', 0xBFA6),
('L', 0xC0AC), ('M', 0xC2E8), ('N', 0xC4C3), ('O', 0xC5B6), ('P', 0xC5BE),
('Q', 0xC6DA), ('R', 0xC8BB), ('S', 0xC8F6), ('T', 0xCBFA), ('W', 0xCDDA),
('X', 0xCEF4), ('Y', 0xD1B9), ('Z', 0xD4D1),
]
_pinyin_boundaries_end = 0xF7FF
def _char_pinyin_initial(char):
"""返回单个汉字 / 数字 / 字母的拼音首字母,非 GB2312 汉字返回 '#'"""
code = ord(char)
if 0x30 <= code <= 0x39: # 0-9
return str(char)
if 0x41 <= code <= 0x5A: # A-Z
return chr(code)
if 0x61 <= code <= 0x7A: # a-z
return chr(code - 32)
if 0x4E00 <= code <= 0x9FA5: # GBK 汉字范围(含 GB2312
try:
gb = char.encode('gb2312')
except UnicodeEncodeError:
return '#'
low = (gb[0] << 8) | gb[1]
if low < 0xB0A1 or low >= _pinyin_boundaries_end:
return '#'
for letter, start in reversed(_PYINYIN_BOUNDARIES):
if low >= start:
return letter
return '#'
def _username_pinyin_abbr(username):
"""返回用户名各字符的拼音首字母串('张伟恒' -> 'ZWH'),供通讯录分组与模糊搜索使用。"""
name = (username or '').strip()
if not name:
return '#'
return ''.join(_char_pinyin_initial(ch) for ch in name)
def _internal_error(exc, status=500):
"""记录内部错误日志,仅向客户端返回通用消息,避免泄露内部信息。"""
current_app.logger.error('[admin_api] internal error: %s', exc, exc_info=True)
@@ -107,6 +151,11 @@ ADMIN_MENU_ACCESS_CONFIG = {
'route_path': 'shop-data-crawl-tasks',
'error': '无权访问店铺数据记录模块',
},
'shop-data-duplicate-check': {
'column_key': 'admin_shop_data_duplicate_check',
'route_path': 'shop-data-duplicate-check',
'error': '无权访问店铺数据重复检查模块',
},
}
ADMIN_MENU_ACCESS_CONFIG.update({
@@ -183,8 +232,38 @@ def _backend_java_internal_headers():
return headers
def _resolve_system_operator_id():
"""无请求上下文(定时扫描线程/脚本)时,取一个系统级超级管理员作为内部操作员。
Java 内部端点(如结果下载)要求 operatorId 指向真实存在的管理员;
传 0 会被 Java 判定「用户不存在」并返回 JSON 错误体。
优先超管(内部端点对超管直接放行);兜底任一管理员。
"""
conn = get_db()
try:
with conn.cursor() as cur:
cur.execute(
"SELECT id FROM users WHERE role = 'super_admin' "
"OR (is_admin = 1 AND created_by_id IS NULL) ORDER BY id LIMIT 1")
row = cur.fetchone()
if row and row.get('id'):
return int(row['id'])
cur.execute(
"SELECT id FROM users WHERE role = 'admin' OR is_admin = 1 "
"ORDER BY id LIMIT 1")
row = cur.fetchone()
if row and row.get('id'):
return int(row['id'])
finally:
conn.close()
return None
def _backend_java_internal_request():
headers = _backend_java_internal_headers()
# 无请求上下文(定时扫描线程/脚本)时没有当前管理员,
# 用系统级超级管理员代替内部操作员(传 0 会被 Java 端判定「用户不存在」)
if has_request_context():
_, current_row = get_current_admin_role()
operator_id = _get_current_admin_id(current_row)
try:
@@ -193,6 +272,10 @@ def _backend_java_internal_request():
raise ValueError('当前管理员身份无效') from exc
if operator_id <= 0:
raise ValueError('当前管理员身份无效')
else:
operator_id = _resolve_system_operator_id()
if not operator_id:
raise ValueError('系统管理员不存在,无法调起内部接口')
return headers, {'operatorId': operator_id}
@@ -824,6 +907,55 @@ def _ensure_shop_data_crawl_data_access():
return role, current_row, (jsonify({'success': False, 'error': '无权查看店铺数据任务'}), 403)
def _shop_data_managed_groups(role, current_row):
"""当前主管可管理的分组 id 列表(超管返回 None 表示不限)。"""
if role == 'super_admin':
return None
admin_id = int(current_row['id']) if current_row and current_row.get('id') else 0
if not admin_id:
return []
conn = get_db()
try:
with conn.cursor() as cur:
cur.execute(
'SELECT id FROM biz_shop_manage_group '
'WHERE created_by_id = %s OR user_id = %s',
(admin_id, admin_id),
)
group_ids = [int(row['id']) for row in cur.fetchall() if row.get('id')]
finally:
conn.close()
return group_ids
def _shop_data_managed_shop_names(role, current_row):
"""当前主管可见的店铺名集合(含自身创建/所属组下的店铺)。
超管返回 None 表示全部;主管返回小写归一后的名字集合,用于过滤缓存明细。
"""
group_ids = _shop_data_managed_groups(role, current_row)
if group_ids is None:
return None
if not group_ids:
return set()
conn = get_db()
try:
with conn.cursor() as cur:
placeholders = ','.join(['%s'] * len(group_ids))
cur.execute(
f'SELECT TRIM(sm.shop_name) AS shop_name FROM biz_shop_manage sm WHERE sm.group_id IN ({placeholders})',
tuple(group_ids),
)
shop_names = {
_shop_data_crawl_shop_key(row.get('shop_name'))
for row in cur.fetchall()
if row.get('shop_name')
}
finally:
conn.close()
return shop_names
def _ensure_product_category_access():
role, current_row, items, denied = _load_current_backend_menu_items()
if denied:
@@ -1020,6 +1152,7 @@ def list_users():
'created_by_id': r.get('created_by_id'),
'creator_username': r.get('creator_username') or '',
'created_at': r['created_at'].strftime('%Y-%m-%d %H:%M') if r.get('created_at') else '',
'pinyin_abbr': _username_pinyin_abbr(r.get('username')),
}
for r in rows
]
@@ -1696,6 +1829,109 @@ def _load_shop_data_crawl_download_rows(result_ids):
_SHOP_DATA_CRAWL_ASIN_ANALYSIS_MAX_BYTES = 256 * 1024 * 1024
_SHOP_DATA_CRAWL_ASIN_ANALYSIS_TIMEOUT = (10, 60)
# 重复 ASIN 扫描缓存:结果存 MySQLshop_data_duplicate_scan 表),
# 定时任务每天凌晨执行一次全量扫描,页面读取最近一次结果,避免实时拉取 Excel 消耗资源。
_DUPLICATE_SCAN_TABLE = 'shop_data_duplicate_scan'
_duplicate_scan_lock = threading.Lock()
_duplicate_scan_running = False
def _ensure_duplicate_scan_table():
"""按需建表:重复 ASIN 扫描结果缓存表。"""
conn = get_db()
try:
with conn.cursor() as cur:
cur.execute(
f"""
CREATE TABLE IF NOT EXISTS {_DUPLICATE_SCAN_TABLE} (
id BIGINT AUTO_INCREMENT PRIMARY KEY,
status VARCHAR(16) NOT NULL,
error_text TEXT NULL,
summary_json MEDIUMTEXT NULL,
payload_json MEDIUMTEXT NULL,
created_at DATETIME NOT NULL DEFAULT CURRENT_TIMESTAMP,
finished_at DATETIME NULL,
KEY idx_status_created (status, created_at)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4
"""
)
conn.commit()
finally:
conn.close()
def _save_duplicate_scan(status, summary=None, payload=None, error_text=None):
"""保存一次扫描结果记录;status: SUCCESS / FAILED。"""
import json as _json
conn = get_db()
scan_id = None
try:
with conn.cursor() as cur:
cur.execute(
f'INSERT INTO {_DUPLICATE_SCAN_TABLE} (status, error_text, summary_json, payload_json, created_at, finished_at) '
'VALUES (%s, %s, %s, %s, NOW(), NOW())',
(
status,
error_text,
_json.dumps(summary or {}, ensure_ascii=False),
_json.dumps(payload or [], ensure_ascii=False),
),
)
scan_id = cur.lastrowid
conn.commit()
return scan_id
finally:
conn.close()
def _latest_duplicate_scan():
"""最近一次成功扫描记录:{'scanned_at': str, 'summary': dict, 'items': list, 'shops': list}|None。
兼容两种缓存格式:
- 新版(重复检查页):payload = {'shops': [...], 'items': [...]}
- 旧版(跨店重复明细):payload 直接为 items 列表,shops 由明细推导。
"""
import json as _json
conn = get_db()
try:
with conn.cursor() as cur:
cur.execute(
f'SELECT id, summary_json, payload_json, created_at FROM {_DUPLICATE_SCAN_TABLE} '
"WHERE status = 'SUCCESS' ORDER BY id DESC LIMIT 1"
)
row = cur.fetchone()
finally:
conn.close()
if not row:
return None
summary_raw = row.get('summary_json') or {}
payload = row.get('payload_json') or '[]'
# summary_json / payload_json 在 MySQL 中为 JSON 字符串,读取后需反序列化
summary = _json.loads(summary_raw) if isinstance(summary_raw, str) else (summary_raw or {})
summary = summary if isinstance(summary, dict) else {}
payload = _json.loads(payload) if isinstance(payload, str) else (payload or [])
if isinstance(payload, dict):
shops = payload.get('shops') or []
items = payload.get('items') or []
else:
# 旧格式缓存:兼容旧页面,店铺概览由明细推导
items = payload if isinstance(payload, list) else []
shop_seen = {}
for item in items:
for occ in (item.get('occurrences') or []):
shop_name = occ.get('shop_name') or ''
shop_seen[shop_name] = shop_seen.get(shop_name, 0) + 1
shops = [
{'shop_name': shop_name, 'asin_count': 0, 'group_name': ''}
for shop_name in sorted(shop_seen, key=lambda name: (-shop_seen[name], name))
]
return {
'scanned_at': str(row.get('created_at') or ''),
'summary': summary,
'items': items if isinstance(items, list) else [],
'shops': shops if isinstance(shops, list) else [],
}
def _shop_data_crawl_fetch_result_bytes(row, timeout=None):
"""从 Java 下载接口拉取结果文件字节流(仅内存,不落盘)。"""
@@ -1721,7 +1957,18 @@ def _shop_data_crawl_fetch_result_bytes(row, timeout=None):
if total > _SHOP_DATA_CRAWL_ASIN_ANALYSIS_MAX_BYTES:
raise ValueError('结果文件过大,无法分析')
chunks.append(chunk)
return b''.join(chunks)
raw = b''.join(chunks)
if not raw:
raise ValueError(f'结果文件为空 result_id={result_id}')
# Java 内部端点鉴权/归属失败时返回 JSON 错误体(HTTP 200),解析报错更直观
if raw.lstrip().startswith(b'{'):
try:
err_payload = json.loads(raw.decode('utf-8', 'replace'))
message = err_payload.get('message') or err_payload.get('error') or '未知错误'
except ValueError:
message = raw[:200].decode('utf-8', 'replace')
raise ValueError(f'结果文件下载失败 result_id={result_id}: {message}')
return raw
finally:
response.close()
@@ -1763,8 +2010,26 @@ def _shop_data_date_key(date_text):
return text
_SHOP_DATA_SHEET_COUNTRY_MAP = {
'英国': 'UK', 'uk': 'UK', 'u.k.': 'UK', 'united kingdom': 'UK',
'德国': 'DE', 'de': 'DE', 'germany': 'DE',
'法国': 'FR', 'fr': 'FR', 'france': 'FR',
'意大利': 'IT', 'it': 'IT', 'italy': 'IT',
'西班牙': 'ES', 'es': 'ES', 'spain': 'ES',
}
def _shop_data_sheet_country(sheet_name):
"""结果文件 sheet 名 → 国家码(英国/德国/法国…);无法识别返回空串。"""
key = str(sheet_name or '').strip().lower()
return _SHOP_DATA_SHEET_COUNTRY_MAP.get(key, '')
def _shop_data_crawl_parse_workbook(workbook):
"""从结果 Workbook 提取 ASIN 行:字典列表,表头按 日期/ASIN/…/价格/…/品牌 识别,跳过无 ASIN 行。"""
"""从结果 Workbook 提取 ASIN 行:字典列表,表头按 日期/ASIN/…/价格/…/品牌 识别,跳过无 ASIN 行。
每行附带所在 sheet 映射的国家码 country(记录级站点,供重复检查页按站点筛选/展示)。
"""
rows = []
for sheet in workbook.worksheets:
header_cells = list(next(sheet.iter_rows(min_row=1, max_row=1), []))
@@ -1776,6 +2041,7 @@ def _shop_data_crawl_parse_workbook(workbook):
date_col = header.index('日期') if '日期' in header else None
price_col = header.index('价格') if '价格' in header else None
brand_col = header.index('品牌') if '品牌' in header else None
sheet_country = _shop_data_sheet_country(sheet.title)
for sheet_row in sheet.iter_rows(min_row=2):
asin = _shop_data_crawl_cell_text(sheet_row[asin_col])
if not asin:
@@ -1785,42 +2051,19 @@ def _shop_data_crawl_parse_workbook(workbook):
'date': _shop_data_crawl_cell_text(sheet_row[date_col]) if date_col is not None else '',
'price': _shop_data_crawl_cell_text(sheet_row[price_col]) if price_col is not None else '',
'brand': _shop_data_crawl_cell_text(sheet_row[brand_col]) if brand_col is not None else '',
'country': sheet_country,
})
return rows
@admin_api.route('/shop-data-crawl/duplicate-asins')
@login_required
def shop_data_crawl_duplicate_asins():
"""按当前店铺列表筛选条件,分析跨店铺重复的 ASIN 明细。
读取当前页每家店铺最新结果 Excel(经 Java 下载接口拉取),按 ASIN 聚合
其出现的店铺、国家与「日期/价格/品牌」细节,仅返回出现在 2 家及以上店铺
的 ASIN(按店铺数降序、ASIN 升序),空 pagination 参数时返回全量用于导出。
"""
_, _, denied = _ensure_backend_menu_access('shop-data-crawl-tasks')
if not denied:
_, _, denied = _ensure_shop_data_crawl_data_access()
if denied:
return denied
try:
page = max(1, int(request.args.get('page', 1)))
page_size = min(100, max(10, int(request.args.get('page_size', 20))))
shop_name_filter = (request.args.get('shop_name') or request.args.get('shop') or '').strip()
country_filter = (request.args.get('country') or '').strip().upper()
asin_filter = (request.args.get('asin') or '').strip().upper()
date_from = (request.args.get('date_from') or '').strip()[:10]
date_to = (request.args.get('date_to') or '').strip()[:10]
def _shops_with_latest_results():
"""查询每家店铺最新结果行(用于重复 ASIN 扫描)。返回 shop_items 列表。"""
conditions = [
"r.module_type = 'SHOP_DATA_CRAWL'",
"t.module_type = 'SHOP_DATA_CRAWL'",
"TRIM(COALESCE(r.result_file_url, '')) <> ''",
]
params = []
where_sql = ' AND '.join(conditions)
offset = (page - 1) * page_size
conn = get_db()
try:
with conn.cursor() as cur:
@@ -1832,26 +2075,15 @@ def shop_data_crawl_duplicate_asins():
'LEFT JOIN users u ON u.id = r.user_id '
'WHERE ' + where_sql
)
cur.execute(
'SELECT COUNT(*) AS total FROM ('
'SELECT ' + shop_key_sql + ' AS shop_key' + grouped_from_sql +
' GROUP BY ' + shop_key_sql +
') shop_groups',
tuple(params),
)
total = int((cur.fetchone() or {}).get('total') or 0)
cur.execute(
'SELECT ' + shop_key_sql + ' AS shop_name, MAX('
+ _SHOP_DATA_CRAWL_LATEST_TIME_SQL + ') AS latest_created_at'
+ grouped_from_sql +
' GROUP BY ' + shop_key_sql +
' ORDER BY latest_created_at DESC, shop_name ASC LIMIT %s OFFSET %s',
tuple(params + [page_size, offset]),
' ORDER BY latest_created_at DESC, shop_name ASC',
)
group_rows = cur.fetchall()
group_names = _shop_data_crawl_group_names(cur, group_rows)
result_rows_by_shop = {}
selected_shop_names = [row.get('shop_name') for row in group_rows]
if selected_shop_names:
@@ -1869,7 +2101,7 @@ def shop_data_crawl_duplicate_asins():
f' AND {shop_key_sql} IN ({placeholders})' +
') ranked WHERE ranked.shop_row_number <= 1 '
'ORDER BY ranked.latest_file_updated_at DESC, ranked.result_id DESC',
tuple(params + selected_shop_names),
tuple(selected_shop_names),
)
for row in cur.fetchall():
shop_key = _shop_data_crawl_shop_key(row.get('shop_name'))
@@ -1877,7 +2109,7 @@ def shop_data_crawl_duplicate_asins():
finally:
conn.close()
# 逐店读取结果文件并解析(每店最多一个结果文件,共 page_size 个)
# 逐店读取结果文件并解析
shop_items = []
for rows in result_rows_by_shop.values():
for row in rows:
@@ -1889,7 +2121,7 @@ def shop_data_crawl_duplicate_asins():
try:
parsed = _shop_data_crawl_parse_workbook(
load_workbook(io.BytesIO(raw), read_only=True, data_only=True))
except (InvalidFileException, KeyError, ValueError) as exc:
except (InvalidFileException, KeyError, ValueError, zipfile.BadZipFile) as exc:
current_app.logger.warning(
'[shop-data-crawl] 解析结果文件失败 result_id=%s: %s', result_id, exc)
continue
@@ -1903,14 +2135,22 @@ def shop_data_crawl_duplicate_asins():
except (requests.RequestException, ValueError) as exc:
current_app.logger.warning(
'[shop-data-crawl] 拉取结果文件失败 result_id=%s: %s', result_id, exc)
return shop_items
# 先全量聚合所有行,再用筛选条件圈定「命中 ASIN」;
# 命中且跨店重复的 ASIN 展示完整 occurrences(方便看该 ASIN 与哪些店重复)。
def _build_duplicate_scan_cache(shop_items, source='job'):
"""由 shop_items 构建全量重复检查缓存:shops 概览 + 明细 items + 指标统计。
返回 {'shops': [...], 'items': [...], 'summary': {唯一ASIN/上架记录/重复ASIN…}}
items 内 occurrences 记录带 country(结果文件 sheet 映射的站点)。
"""
asin_occurrences = {}
matched_asins = set()
shop_agg = {}
for shop_item in shop_items:
shop_name = shop_item['shop_name'] or '未命名'
group_name = shop_item['group_name'] or ''
country_codes = shop_item['country_codes'] or []
shop_agg[shop_name] = {'group_name': group_name, 'country_codes': country_codes}
for row in shop_item['rows']:
asin = row['asin'].strip().upper()
if not asin:
@@ -1921,40 +2161,445 @@ def shop_data_crawl_duplicate_asins():
'price': row['price'],
'brand': row['brand'],
'shop_name': shop_name,
'group_name': shop_item['group_name'],
'country_codes': shop_item['country_codes'],
'group_name': group_name,
'country_codes': country_codes,
'country': row.get('country') or '',
})
# 行级筛选:命中则将此 ASIN 加入待展示集合
if asin_filter and asin_filter not in asin:
continue
if shop_name_filter and shop_name_filter.lower() not in (shop_name or '').lower() \
and shop_name_filter.lower() not in (shop_item['group_name'] or '').lower():
continue
if country_filter and country_filter not in [c.upper() for c in country_codes]:
continue
row_date_key = _shop_data_date_key((row['date'] or '').strip())
if date_from and row_date_key and row_date_key < date_from:
continue
if date_to and row_date_key and row_date_key > date_to:
continue
matched_asins.add(asin)
occurrences_list = []
for asin in matched_asins:
occurrences = asin_occurrences.get(asin, [])
items = []
total_records = 0
for asin, occurrences in asin_occurrences.items():
shop_count = len({item['shop_name'] for item in occurrences})
if shop_count < 2:
continue
occurrences_list.append({
total_records += len(occurrences)
items.append({
'asin': asin,
'shop_count': shop_count,
'record_count': len(occurrences),
'occurrences': occurrences,
})
occurrences_list.sort(key=lambda item: (-item['shop_count'], item['asin']))
items.sort(key=lambda item: (-item['shop_count'], item['asin']))
repeated = sum(1 for item in items if item['shop_count'] >= 2)
total_details = len(occurrences_list)
paged_details = occurrences_list[offset:offset + page_size]
shops = []
for shop_name, agg in sorted(shop_agg.items()):
shop_asin_set = {
occ['asin']
for occ_list in (oo for oo in asin_occurrences.values())
for occ in occ_list
if occ['shop_name'] == shop_name
}
shops.append({
'shop_name': shop_name,
'group_name': agg['group_name'],
'country_codes': agg['country_codes'],
'asin_count': len(shop_asin_set),
'record_count': sum(1 for occ_list in asin_occurrences.values() for occ in occ_list if occ['shop_name'] == shop_name),
})
shops.sort(key=lambda shop: (-shop['asin_count'], shop['shop_name']))
summary = {
'shop_count': len(shops),
'asin_total': len(items),
'record_total': total_records,
'duplicate_asin_total': repeated,
'duplicate_shop_count': len({occ['shop_name'] for item in items if item['shop_count'] >= 2 for occ in item['occurrences']}),
'site_count': len({occ['country'] for occ_list in asin_occurrences.values() for occ in occ_list if occ.get('country')}),
'asin_per_shop': round(float(total_records) / len(shops), 1) if shops else 0.0,
'source': source,
}
return {'shops': shops, 'items': items, 'summary': summary}
def _build_duplicate_scan_items(shop_items):
"""由 shop_items 聚合跨店重复 ASIN 完整结果(不含筛选,返回全量列表)。
保持旧接口语义(仅跨店重复),供导入模拟重分析等旧逻辑复用;
新页面走 _build_duplicate_scan_cache 全量缓存。
"""
cache = _build_duplicate_scan_cache(shop_items)
return cache['items']
def _run_duplicate_scan_job():
"""执行一次全量重复检查扫描并落库(新格式:shops + items + summary)。返回 (ok, scanned_at, summary)。"""
global _duplicate_scan_running
if _duplicate_scan_running:
return False, '', '扫描进行中'
_duplicate_scan_running = True
try:
shop_items = _shops_with_latest_results()
cache = _build_duplicate_scan_cache(shop_items, source='job')
_save_duplicate_scan('SUCCESS', summary=cache['summary'], payload={
'shops': cache['shops'],
'items': cache['items'],
})
return True, datetime.now().strftime('%Y-%m-%d %H:%M:%S'), cache['summary']
except Exception as exc:
current_app.logger.exception('[shop-data-crawl] 重复检查定时扫描失败: %s', exc)
try:
_save_duplicate_scan('FAILED', error_text=str(exc))
except Exception:
current_app.logger.exception('[shop-data-crawl] 重复检查扫描失败记录落库失败')
return False, '', '执行失败:%s' % exc
finally:
_duplicate_scan_running = False
def start_duplicate_scan_scheduler(app):
"""每天凌晨 03:10 执行一次重复 ASIN 全量扫描(后台守护线程)。"""
_ensure_duplicate_scan_table()
def _loop():
while True:
now = datetime.now()
# 下一个 03:10(含今天,若已过则明天)
next_run = now.replace(hour=3, minute=10, second=0, microsecond=0)
if next_run <= now:
next_run = next_run.replace(day=next_run.day) + timedelta(days=1)
delay = (next_run - now).total_seconds()
app.logger.info('[shop-data-crawl] 重复ASIN定时扫描将于 %s 执行(%d 秒后)', next_run, int(delay))
time.sleep(delay)
try:
with app.app_context():
ok, scanned_at, summary = _run_duplicate_scan_job()
app.logger.info('[shop-data-crawl] 重复ASIN定时扫描结束 ok=%s summary=%s', ok, summary)
except Exception as exc:
app.logger.exception('[shop-data-crawl] 重复ASIN定时扫描异常: %s', exc)
threading.Thread(target=_loop, daemon=True, name='duplicate-scan-scheduler').start()
def _duplicate_check_filter_cache(cache, role, current_row):
"""按角色数据范围裁剪重复检查缓存:超管返回全量,主管只返回自己组的店铺。
返回 (shops, items)shops 为可见店铺概览;items 为裁剪 occurrences 后的明细
(裁剪后同一 ASIN 只剩 1 家店时也会保留 —— 展示完整台账,是否重复由 shop_count>=2 标记)。
"""
visible_shops = _shop_data_managed_shop_names(role, current_row)
if visible_shops is None:
return cache.get('shops') or [], cache.get('items') or []
shops = []
for shop in (cache.get('shops') or []):
if _shop_data_crawl_shop_key(shop.get('shop_name')) in visible_shops:
shops.append(shop)
items = []
for item in (cache.get('items') or []):
occurrences = [
occ for occ in (item.get('occurrences') or [])
if _shop_data_crawl_shop_key(occ.get('shop_name')) in visible_shops
]
if occurrences:
items.append({
'asin': item.get('asin'),
'shop_count': len({occ.get('shop_name') for occ in occurrences}),
'record_count': len(occurrences),
'occurrences': occurrences,
})
return shops, items
@admin_api.route('/shop-data-crawl/duplicate-check-overview')
@login_required
def shop_data_crawl_duplicate_check_overview():
"""店铺数据重复检查:指标卡 + 店铺上架分布(按当前角色数据范围裁剪)。"""
role, current_row, denied = _ensure_backend_menu_access('shop-data-crawl-tasks')
if not denied:
_, _, denied = _ensure_shop_data_crawl_data_access()
if denied:
return denied
try:
force = (request.args.get('force') or '').strip() in ('1', 'true', 'yes')
if force:
if not _duplicate_scan_lock.acquire(blocking=False):
return jsonify({'success': False, 'error': '扫描进行中,请稍后刷新'}), 409
try:
ok, scanned_at, summary = _run_duplicate_scan_job()
if not ok:
return jsonify({'success': False, 'error': summary}), 500
cache = _latest_duplicate_scan()
finally:
_duplicate_scan_lock.release()
else:
cache = _latest_duplicate_scan()
if not cache:
return jsonify({
'success': True, 'pending': True,
'scanned_at': '', 'source': '',
'summary': {}, 'shops': [],
})
shops, items = _duplicate_check_filter_cache(cache, role, current_row)
asin_total = len(items)
record_total = sum(item.get('record_count') or 0 for item in items)
repeated = [item for item in items if item['shop_count'] >= 2]
shop_asin_totals = {}
duplicate_shops = set()
site_codes = set()
for item in items:
occs = item.get('occurrences') or []
for occ in occs:
shop_asin_totals[occ.get('shop_name')] = shop_asin_totals.get(occ.get('shop_name'), set())
shop_asin_totals[occ.get('shop_name')].add(item['asin'])
if item['shop_count'] >= 2:
duplicate_shops.add(occ.get('shop_name'))
if occ.get('country'):
site_codes.add(occ['country'])
summary = {
'shop_count': len(shops),
'asin_total': asin_total,
'record_total': record_total,
'duplicate_asin_total': len(repeated),
'duplicate_shop_count': len(duplicate_shops),
'site_count': len(site_codes),
'asin_per_shop': round(float(record_total) / len(shops), 1) if shops else 0.0,
'source': (cache.get('summary') or {}).get('source', ''),
}
return jsonify({
'success': True,
'pending': False,
'scanned_at': cache['scanned_at'],
'summary': summary,
'shops': shops,
})
except Exception as exc:
return _internal_error(exc)
@admin_api.route('/shop-data-crawl/duplicate-check-items')
@login_required
def shop_data_crawl_duplicate_check_items():
"""店铺数据重复检查矩阵/台账明细:行 ASIN、列店铺,支持筛选与分页。"""
_, _, denied = _ensure_backend_menu_access('shop-data-crawl-tasks')
if not denied:
_, _, denied = _ensure_shop_data_crawl_data_access()
if denied:
return denied
try:
role, current_row = get_current_admin_role()
page = max(1, int(request.args.get('page', 1)))
page_size = min(100, max(10, int(request.args.get('page_size', 20))))
view = (request.args.get('view') or 'monitor').strip().lower()
asin_filter = (request.args.get('asin') or '').strip().upper()
shop_name_filter = (request.args.get('shop_name') or request.args.get('shop') or '').strip()
country_filter = (request.args.get('country') or '').strip().upper()
site_filter = (request.args.get('site') or '').strip().upper()
date_from = (request.args.get('date_from') or '').strip()[:10]
date_to = (request.args.get('date_to') or '').strip()[:10]
cache = _latest_duplicate_scan()
if not cache:
return jsonify({
'success': True, 'pending': True,
'items': [], 'shops': [],
'total': 0, 'page': page, 'page_size': page_size,
'scanned_at': '',
})
shops, all_items = _duplicate_check_filter_cache(cache, role, current_row)
has_filter = bool(asin_filter or shop_name_filter or country_filter or site_filter or date_from or date_to)
matched = []
for item in all_items:
if asin_filter and asin_filter not in item['asin']:
continue
if view == 'monitor' and item['shop_count'] < 2:
continue
if not has_filter:
matched.append(item)
continue
hit = False
for occ in item['occurrences']:
if shop_name_filter and shop_name_filter.lower() not in (occ.get('shop_name') or '').lower() \
and shop_name_filter.lower() not in (occ.get('group_name') or '').lower():
continue
if country_filter and country_filter not in [c.upper() for c in (occ.get('country_codes') or [])]:
continue
if site_filter and site_filter not in [c.upper() for c in [occ.get('country') or '']]:
continue
row_date_key = _shop_data_date_key((occ.get('date') or '').strip())
if date_from and row_date_key and row_date_key < date_from:
continue
if date_to and row_date_key and row_date_key > date_to:
continue
hit = True
break
if hit:
matched.append(item)
total = len(matched)
offset = (page - 1) * page_size
return jsonify({
'success': True,
'pending': False,
'items': matched[offset:offset + page_size],
'shops': shops,
'total': total,
'page': page,
'page_size': page_size,
'scanned_at': cache['scanned_at'],
})
except ValueError as exc:
return jsonify({'success': False, 'error': str(exc)}), 400
except Exception as exc:
return _internal_error(exc)
def _duplicate_check_effective_site(country, country_codes):
"""导出用站点:优先记录级 country,其次店铺 country_codes 汇总。"""
if country:
return country.upper()
codes = [c.upper() for c in (country_codes or []) if c]
return ''.join(codes) if codes else ''
@admin_api.route('/shop-data-crawl/duplicate-check-export')
@login_required
def shop_data_crawl_duplicate_check_export():
"""店铺数据重复检查导出 CSV:筛选逻辑与矩阵接口一致,导出全量匹配行。"""
_, _, denied = _ensure_backend_menu_access('shop-data-crawl-tasks')
if not denied:
_, _, denied = _ensure_shop_data_crawl_data_access()
if denied:
return denied
try:
role, current_row = get_current_admin_role()
view = (request.args.get('view') or 'monitor').strip().lower()
asin_filter = (request.args.get('asin') or '').strip().upper()
shop_name_filter = (request.args.get('shop_name') or request.args.get('shop') or '').strip()
country_filter = (request.args.get('country') or '').strip().upper()
site_filter = (request.args.get('site') or '').strip().upper()
date_from = (request.args.get('date_from') or '').strip()[:10]
date_to = (request.args.get('date_to') or '').strip()[:10]
cache = _latest_duplicate_scan()
if not cache:
return jsonify({'success': False, 'error': '暂无扫描结果,请先点击「重新分析」'}), 400
shops, all_items = _duplicate_check_filter_cache(cache, role, current_row)
has_filter = bool(asin_filter or shop_name_filter or country_filter or site_filter or date_from or date_to)
# 展平为逐行记录(每行=一条上架记录),筛选逻辑与矩阵接口一致
flat = []
for item in all_items:
if asin_filter and asin_filter not in item['asin']:
continue
if view == 'monitor' and item['shop_count'] < 2:
continue
for occ in item['occurrences']:
if shop_name_filter and shop_name_filter.lower() not in (occ.get('shop_name') or '').lower() \
and shop_name_filter.lower() not in (occ.get('group_name') or '').lower():
continue
if country_filter and country_filter not in [c.upper() for c in (occ.get('country_codes') or [])]:
continue
if site_filter and site_filter not in [c.upper() for c in [occ.get('country') or '']]:
continue
row_date_key = _shop_data_date_key((occ.get('date') or '').strip())
if date_from and row_date_key and row_date_key < date_from:
continue
if date_to and row_date_key and row_date_key > date_to:
continue
flat.append({
'asin': item['asin'],
'shop_count': item['shop_count'],
'shop_name': occ.get('shop_name') or '',
'group_name': occ.get('group_name') or '',
'country': _duplicate_check_effective_site(occ.get('country'), occ.get('country_codes')),
'date': occ.get('date') or '',
'price': occ.get('price') or '',
'brand': occ.get('brand') or '',
})
flat.sort(key=lambda r: (-r['shop_count'], r['asin'], r['shop_name'], r['date']))
import csv
stream = io.StringIO()
writer = csv.writer(stream)
writer.writerow(['ASIN', '店铺数', '店铺', '分组', '站点', '上架时间', '价格', '品牌'])
for row in flat:
writer.writerow([row['asin'], row['shop_count'], row['shop_name'], row['group_name'],
row['country'], row['date'], row['price'], row['brand']])
payload = ('' + stream.getvalue()).encode('utf-8')
response = Response(payload, mimetype='text/csv; charset=utf-8')
response.headers['Content-Disposition'] = 'attachment; filename="shop-data-duplicate-check.csv"'
response.headers['Cache-Control'] = 'no-store'
return response
except Exception as exc:
return _internal_error(exc)
@admin_api.route('/shop-data-crawl/duplicate-asins')
@login_required
def shop_data_crawl_duplicate_asins():
"""按当前筛选条件展示跨店铺重复的 ASIN 明细。
默认读取最近一次成功扫描的缓存结果(定时任务每天凌晨全量扫描),
页面点「重新分析」携带 force=1 触发一次实时扫描(锁防并发,重复触发返回 409)。
"""
_, _, denied = _ensure_backend_menu_access('shop-data-crawl-tasks')
if not denied:
_, _, denied = _ensure_shop_data_crawl_data_access()
if denied:
return denied
try:
page = max(1, int(request.args.get('page', 1)))
page_size = min(100, max(10, int(request.args.get('page_size', 20))))
force = (request.args.get('force') or '').strip() in ('1', 'true', 'yes')
shop_name_filter = (request.args.get('shop_name') or request.args.get('shop') or '').strip()
country_filter = (request.args.get('country') or '').strip().upper()
asin_filter = (request.args.get('asin') or '').strip().upper()
date_from = (request.args.get('date_from') or '').strip()[:10]
date_to = (request.args.get('date_to') or '').strip()[:10]
if force:
# 实时扫描:持锁执行,避免与定时任务/其他请求并发
if not _duplicate_scan_lock.acquire(blocking=False):
return jsonify({'success': False, 'error': '扫描进行中,请稍后刷新'}), 409
try:
ok, scanned_at, summary = _run_duplicate_scan_job()
if not ok:
return jsonify({'success': False, 'error': summary}), 500
cache = _latest_duplicate_scan()
finally:
_duplicate_scan_lock.release()
else:
cache = _latest_duplicate_scan()
if not cache:
return jsonify({
'success': True,
'items': [],
'total': 0,
'page': page,
'page_size': page_size,
'analyzed_shop_count': 0,
'analyzed_result_count': 0,
'scanned_at': '',
'pending': True,
})
all_items = cache['items'] or []
# 旧接口只展示跨店重复:由全量缓存裁剪出 shop_count>=2 的 ASIN
all_items = [item for item in all_items if (item.get('shop_count') or 0) >= 2]
has_filter = bool(asin_filter or shop_name_filter or country_filter or date_from or date_to)
matched = []
for item in all_items:
if asin_filter and asin_filter not in item['asin']:
continue
if not has_filter:
matched.append(item)
continue
# 有筛选时:任一 occurrence 命中即保留该 ASIN(展示完整记录)
hit = False
for occ in item['occurrences']:
if shop_name_filter and shop_name_filter.lower() not in (occ.get('shop_name') or '').lower() \
and shop_name_filter.lower() not in (occ.get('group_name') or '').lower():
continue
if country_filter and country_filter not in [c.upper() for c in (occ.get('country_codes') or [])]:
continue
row_date_key = _shop_data_date_key((occ.get('date') or '').strip())
if date_from and row_date_key and row_date_key < date_from:
continue
if date_to and row_date_key and row_date_key > date_to:
continue
hit = True
break
if hit:
matched.append(item)
total_details = len(matched)
offset = (page - 1) * page_size
paged_details = matched[offset:offset + page_size]
return jsonify({
'success': True,
@@ -1962,8 +2607,10 @@ def shop_data_crawl_duplicate_asins():
'total': total_details,
'page': page,
'page_size': page_size,
'analyzed_shop_count': len(shop_items),
'analyzed_result_count': len(shop_items),
'analyzed_shop_count': cache['summary'].get('shop_count', 0) if cache.get('summary') else 0,
'analyzed_result_count': len(all_items),
'scanned_at': cache['scanned_at'],
'source': cache.get('summary', {}).get('source', ''),
})
except ValueError as exc:
return jsonify({'success': False, 'error': str(exc)}), 400
+571 -196
View File
@@ -97,7 +97,7 @@
var ADMIN_MENU_GROUPS = [
{ key: 'account', title: '账号与权限', items: ['users', 'columns', 'group-manage'] },
{ key: 'data', title: '数据管理', items: ['dedupe-total-data', 'invalid-asin-data', 'query-asin', 'product-categories'] },
{ key: 'shop', title: '店铺管理', items: ['shop-keys', 'shop-manage', 'skip-price-asin', 'shop-data-crawl-tasks'] },
{ key: 'shop', title: '店铺管理', items: ['shop-keys', 'shop-manage', 'skip-price-asin', 'shop-data-crawl-tasks', 'shop-data-duplicate-check'] },
{ key: 'record', title: '记录与版本', items: ['history', 'version', 'digital-human-version', 'image-video-tasks'] }
];
var ADMIN_MENU_ICONS = {
@@ -113,6 +113,7 @@
'product-categories': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M20 20a2 2 0 0 0 2-2V8a2 2 0 0 0-2-2h-7.9a2 2 0 0 1-1.69-.9L9.6 3.9A2 2 0 0 0 7.93 3H4a2 2 0 0 0-2 2v13a2 2 0 0 0 2 2Z"></path></svg>',
'image-video-tasks': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="m22 8-6 4 6 4V8Z"></path><rect width="14" height="12" x="2" y="6" rx="2" ry="2"></rect></svg>',
'shop-data-crawl-tasks': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><ellipse cx="12" cy="5" rx="9" ry="3"></ellipse><path d="M3 5v14a9 3 0 0 0 18 0V5"></path><path d="M3 12a9 3 0 0 0 18 0"></path></svg>',
'shop-data-duplicate-check': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M21.35 11.1h-9.17a2 2 0 0 1-1.75-2.98l1.67-2.79a2 2 0 0 0-1.75-2.98l-5.98-.01a2 2 0 0 0-2 2v1.5a2 2 0 0 0 2 2h3.36l-2.38 3.97a2 2 0 0 0 1.75 2.98h11.5a2 2 0 0 0 2-2v-1.5a2 2 0 0 0-2-2Z"></path><path d="M3 21h18"></path></svg>',
'history': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M3 12a9 9 0 1 0 9-9 9.75 9.75 0 0 0-6.74 2.74L3 8"></path><path d="M3 3v5h5"></path><path d="M12 7v5l4 2"></path></svg>',
'version': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="m7.5 4.27 9 5.15"></path><path d="M21 8a2 2 0 0 0-1-1.73l-7-4a2 2 0 0 0-2 0l-7 4A2 2 0 0 0 3 8v8a2 2 0 0 0 1 1.73l7 4a2 2 0 0 0 2 0l7-4A2 2 0 0 0 21 16Z"></path><path d="M3.3 7 12 12l8.7-5"></path><path d="M12 22V12"></path></svg>',
'digital-human-version': '<svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="1.8" stroke-linecap="round" stroke-linejoin="round" aria-hidden="true"><path d="M12 8V4H8"></path><rect width="16" height="12" x="4" y="8" rx="2"></rect><path d="M2 14h2"></path><path d="M20 14h2"></path><path d="M15 13v2"></path><path d="M9 13v2"></path></svg>'
@@ -139,6 +140,7 @@
'product-categories': 'panel-product-categories',
'image-video-tasks': 'panel-image-video-tasks',
'shop-data-crawl-tasks': 'panel-shop-data-crawl-tasks',
'shop-data-duplicate-check': 'panel-shop-data-duplicate-check',
'history': 'panel-history',
'version': 'panel-version',
'digital-human-version': 'panel-digital-human-version'
@@ -162,6 +164,7 @@
else if (tabName === 'product-categories') loadProductCategories();
else if (tabName === 'image-video-tasks') loadImageVideoTasks(1);
else if (tabName === 'shop-data-crawl-tasks') loadShopDataCrawlTasks(1);
else if (tabName === 'shop-data-duplicate-check') loadShopDataDuplicateCheckOverview();
else if (tabName === 'history') loadHistory(1);
else if (tabName === 'version') loadSoftwareVersions();
else if (tabName === 'digital-human-version') loadDigitalHumanVersions();
@@ -1778,22 +1781,45 @@
}
function renderShopDataRecordTable() {
var rows = shopDataTasks.map(renderShopDataRecordRow).join('');
if (!shopDataTasks.length) {
return '<div class="shop-data-empty-hint">暂无符合条件的店铺数据任务</div>';
}
return '<table>' +
'<thead><tr>' +
'<th style="width:36px;"><input type="checkbox" id="shopDataTaskSelectAll"></th>' +
'<th style="width:18%;">店铺</th>' +
'<th style="width:14%;">分组</th>' +
'<th style="width:16%;">国家</th>' +
'<th style="width:10%;">状态</th>' +
'<th style="width:18%;">更新时间</th>' +
'<th style="width:170px;">操作</th>' +
'</tr></thead>' +
'<tbody>' + rows + '</tbody>' +
'</table>';
// 卡片网格:每店铺一张卡片,按截图样式(店铺名、当日累计文件、分组、最新时间、任务号+状态、国家、文件名、下载/删除)
var cards = shopDataTasks.map(function (item) {
var resultId = shopDataResultId(item);
var selected = resultId > 0 && selectedShopDataResultIds.has(resultId);
var status = String(item.status || item.file_status || '').toUpperCase();
var terminal = ['SUCCESS', 'FAILED', 'CANCELLED'].indexOf(status) >= 0;
var countryCodes = item.country_codes != null ? item.country_codes : item.countryCodes;
var updatedAt = item.updated_at || item.latest_created_at || item.created_at || item.finished_at || '-';
var taskNo = item.task_id || item.taskId || '-';
var errorTitle = item.error ? ' title="' + escapeHtml(item.error) + '"' : '';
return '<article class="shop-data-card" data-shop-data-card="' + (resultId || '') + '">' +
'<div class="shop-data-card-head">' +
'<span class="shop-data-card-shop">' + escapeHtml(item.shop_name || '-') + '</span>' +
'</div>' +
'<div class="shop-data-card-select-row">' +
'<input type="checkbox" data-shop-data-select="' + (resultId || '') + '"' +
(selected ? ' checked' : '') + (item.file_ready && resultId > 0 ? '' : ' disabled') + '>' +
'<span class="shop-data-card-task-no">任务 ' + escapeHtml(String(taskNo)) + '</span>' +
'<span class="shop-data-status ' + (terminal ? (status === 'SUCCESS' ? 'success' : 'failed') : 'running') + '"' + errorTitle + '>' +
escapeHtml(imageVideoStatusLabel(status)) + '</span>' +
'</div>' +
'<div class="shop-data-card-row"><span class="shop-data-card-label">分组</span>' +
'<span class="shop-data-card-value">' + escapeHtml(item.group_name || '-') + '</span></div>' +
'<div class="shop-data-card-row"><span class="shop-data-card-label">最新</span>' +
'<span class="shop-data-card-value">' + escapeHtml(updatedAt) + '</span></div>' +
'<div class="shop-data-card-row"><span class="shop-data-card-label">国家</span>' +
'<span class="shop-data-card-value">' + escapeHtml(countryListLabel(countryCodes)) + '</span></div>' +
'<div class="shop-data-card-row"><span class="shop-data-card-label">文件</span>' +
'<span class="shop-data-card-value shop-data-card-file">' + escapeHtml(item.output_filename || '-') + '</span></div>' +
'<div class="shop-data-card-actions">' +
'<button class="shop-data-record-action" type="button" data-shop-data-download="' + (resultId || '') + '"' + (item.file_ready && resultId > 0 ? '' : ' disabled') + '>' + imageVideoDownloadIcon() + '下载文件</button>' +
'<button class="shop-data-record-action danger" type="button" data-shop-data-delete="' + (resultId || '') + '"' + (terminal && resultId > 0 ? '' : ' disabled') + '>' + shopDataDeleteIcon() + '删除</button>' +
'</div>' +
'</article>';
}).join('');
return '<div class="shop-data-card-grid">' + cards + '</div>';
}
function renderShopDataTasks() {
@@ -1858,21 +1884,27 @@
});
}
// ========== 重复 ASIN 分析 ==========
var shopDataDuplicatePage = 1, shopDataDuplicatePageSize = 10;
// ========== 店铺数据重复检查(指标卡 + 分布 + 矩阵表格 + 抽屉)==========
var shopDataDuplicatePage = 1, shopDataDuplicatePageSize = 20;
var shopDataDuplicateItems = [];
var shopDataDuplicateShops = [];
var shopDataDuplicateTotal = 0;
var shopDataDuplicateAnalyzed = { shopCount: 0, resultCount: 0 };
var shopDataDuplicateView = 'monitor';
var shopDataDuplicateLoading = false;
var shopDataDuplicateScannedAt = '';
var shopDataDuplicateOverviewPending = false;
var shopDataDuplicateExporting = false;
function buildShopDataDuplicateQuery(page) {
var params = new URLSearchParams();
params.set('page', String(page || 1));
params.set('page_size', String(shopDataDuplicatePageSize));
params.set('view', shopDataDuplicateView);
var values = {
asin: document.getElementById('shopDataDupFilterAsin').value.trim(),
shop_name: document.getElementById('shopDataDupFilterShop').value.trim(),
country: document.getElementById('shopDataDupFilterCountry').value.trim(),
site: document.getElementById('shopDataDupFilterSite').value.trim(),
date_from: document.getElementById('shopDataDupFilterDateFrom').value,
date_to: document.getElementById('shopDataDupFilterDateTo').value
};
@@ -1882,166 +1914,372 @@
return params.toString();
}
function shopDataDuplicateHeader(header) {
return '<thead><tr>' + header.map(function (col) {
return '<th style="width:' + (col.width || '') + ';">' + col.label + '</th>';
}).join('') + '</tr></thead>';
function pad2(value) {
var n = parseInt(value, 10);
if (isNaN(n)) return String(value);
return n < 10 ? '0' + n : String(n);
}
function renderShopDataDuplicateCard(item) {
var occurrences = Array.isArray(item.occurrences) ? item.occurrences : [];
var brand = '';
occurrences.forEach(function (occ) { if (occ.brand && !brand) brand = occ.brand; });
var rows = occurrences.map(function (occ) {
var countries = countryListLabel(occ.country_codes);
return '<tr>' +
'<td class="dup-shop">' + escapeHtml(occ.shop_name || '-') + '</td>' +
'<td class="dup-country">' + escapeHtml(occ.group_name || '-') + '</td>' +
'<td class="dup-country">' + escapeHtml(countries) + '</td>' +
'<td class="dup-date">' + escapeHtml(occ.date || '-') + '</td>' +
'<td class="dup-date">' + escapeHtml(occ.price || '-') + '</td>' +
'</tr>';
// 中文日期(如 2026年8月18日 上午4:34)归一化为 YYYY-MM-DD HH:MM[:SS],便于展示与排序
function normalizeDuplicateTime(raw) {
if (!raw) return '';
var s = String(raw).trim();
var m = s.match(/^(\d{4})年(\d{1,2})月(\d{1,2})日\s*(.*)$/);
if (m) {
var md = m[1] + '-' + pad2(m[2]) + '-' + pad2(m[3]);
var tm = m[4].match(/(上午|下午|晚上|凌晨)?\s*(\d{1,2})[:](\d{2})(?:[:](\d{1,2}))?/);
if (!tm) return md;
var hour = parseInt(tm[2], 10);
var period = tm[1] || '';
if (period === '下午' || period === '晚上') hour = hour % 12 + 12;
if ((period === '上午' || period === '凌晨') && hour === 12) hour = 0;
return md + ' ' + pad2(hour) + ':' + tm[3] + (tm[4] ? ':' + pad2(tm[4]) : '');
}
var dm = s.replace(/\//g, '-').match(/^(\d{4})-(\d{1,2})-(\d{1,2})(.*)$/);
if (dm) return dm[1] + '-' + pad2(dm[2]) + '-' + pad2(dm[3]) + (dm[4] || '');
return s;
}
function renderShopDataDuplicateMetrics(summary) {
var metricsEl = document.getElementById('shopDataDuplicateMetrics');
if (!summary || summary.asin_total == null) {
metricsEl.innerHTML = '<div class="dup-check-metric" style="grid-column:1/-1;"><span class="dup-check-metric-label">暂无扫描结果</span>' +
'<span class="dup-check-metric-value">-</span></div>';
return;
}
var defs = [
{ label: '唯一ASIN', value: summary.asin_total, accent: true },
{ label: '上架记录', value: summary.record_total, accent: true },
{ label: '在线店铺', value: summary.shop_count, accent: true },
{ label: '重复ASIN', value: summary.duplicate_asin_total, warn: true },
{ label: '重复店铺', value: summary.duplicate_shop_count, warn: true },
{ label: '店铺平均ASIN', value: summary.asin_per_shop }
];
metricsEl.innerHTML = defs.map(function (def) {
return '<div class="dup-check-metric' + (def.accent ? ' accent' : '') + (def.warn ? ' warn' : '') + '">' +
'<span class="dup-check-metric-label">' + def.label + '</span>' +
'<span class="dup-check-metric-value">' + def.value + '</span></div>';
}).join('');
return '<article class="duplicate-asin-card" data-duplicate-asin="' + escapeHtml(item.asin) + '">' +
'<div class="duplicate-asin-card-head">' +
'<span class="dup-asin">' + escapeHtml(item.asin) + '</span>' +
'<span class="dup-count">' + item.shop_count + ' 家店铺</span>' +
'<span class="dup-brand" title="' + escapeHtml(brand) + '">' + escapeHtml(brand || '') + '</span>' +
'</div>' +
'<table class="duplicate-asin-table">' +
shopDataDuplicateHeader([
{ label: '店铺', width: '18%' },
{ label: '分组', width: '14%' },
{ label: '国家', width: '16%' },
{ label: '日期', width: '14%' },
{ label: '价格', width: '12%' }
]) +
'<tbody>' + rows + '</tbody>' +
'</table>' +
'</article>';
}
function renderShopDataDuplicateDistribution(shops) {
var barsEl = document.getElementById('shopDataDuplicateDistribution');
var noteEl = document.getElementById('shopDataDuplicateDistributionNote');
if (!shops || !shops.length) {
barsEl.innerHTML = '<div class="shop-data-empty-hint">暂无店铺数据</div>';
noteEl.textContent = '';
return;
}
var max = 0;
shops.forEach(function (shop) { if (shop.asin_count > max) max = shop.asin_count; });
barsEl.innerHTML = shops.map(function (shop) {
var width = max ? Math.max(3, Math.round(shop.asin_count / max * 100)) : 0;
return '<div class="dup-chart-row">' +
'<span class="dup-chart-name" title="' + escapeHtml(shop.shop_name) + '">' + escapeHtml(shop.shop_name) + '</span>' +
'<span class="dup-chart-track"><span class="dup-chart-fill" style="width:' + width + '%"></span></span>' +
'<span class="dup-chart-num">' + shop.asin_count + '</span></div>';
}).join('');
noteEl.textContent = '共 ' + shops.length + ' 家店铺 · 按 ASIN 数量排序';
}
function shopDataDuplicateCellCount(item, shopName) {
var count = 0;
(item.occurrences || []).forEach(function (occ) {
if ((occ.shop_name || '') === shopName) count++;
});
return count;
}
function renderShopDataDuplicateMatrix() {
var list = document.getElementById('shopDataDuplicateList');
if (!shopDataDuplicateItems.length) {
list.innerHTML = '<div class="shop-data-empty-hint">暂无数据'
+ (shopDataDuplicateView === 'monitor' && !shopDataDuplicateTotal ? ':当前没有跨店铺重复的 ASIN' : '')
+ '。可点击右上角「重新分析」重新扫描。</div>';
return;
}
var shops = shopDataDuplicateShops;
var thead = '<thead><tr><th>ASIN</th><th class="dup-cell">店铺数</th>' +
shops.map(function (shop) {
return '<th class="dup-cell" title="' + escapeHtml(shop.shop_name) + '">' + escapeHtml(shop.shop_name) + '</th>';
}).join('') + '</tr></thead>';
var tbody = shopDataDuplicateItems.map(function (item) {
var cells = shops.map(function (shop) {
var count = shopDataDuplicateCellCount(item, shop.shop_name);
var cls = count === 0 ? 'zero' : (item.shop_count >= 2 ? 'danger' : 'has');
return '<td class="dup-cell"><span class="dup-cell-num ' + cls + '"'
+ (count ? ' data-open-drawer="' + escapeHtml(item.asin) + '" title="' + escapeHtml(item.asin) + ' 在 ' + escapeHtml(shop.shop_name) + '' + count + ' 条"' : '')
+ '>' + (count === 0 ? '0' : count) + '</span></td>';
}).join('');
return '<tr>' +
'<td class="dup-asin-col" data-open-drawer="' + escapeHtml(item.asin) + '" title="查看详情">' + escapeHtml(item.asin) + '</td>' +
'<td class="dup-cell"><span class="dup-cell-num ' + (item.shop_count >= 2 ? 'danger' : 'has') + '">' + item.shop_count + '</span></td>' +
cells + '</tr>';
}).join('');
list.innerHTML = '<table class="dup-check-table"><thead>' + thead + '</thead><tbody>' + tbody + '</tbody></table>';
}
function renderShopDataDuplicateList() {
var list = document.getElementById('shopDataDuplicateList');
if (!shopDataDuplicateItems.length) {
list.innerHTML = '<div class="shop-data-empty-hint">暂无重复 ASIN。请在筛选条件后点击"查询",或"重新分析"。' +
'<span id="dupEmptyHint"></span></div>';
renderShopDataDuplicateMatrix();
}
// 主时间戳(转数值失败时按字符串比较)
function dupTimeRank(raw) {
var t = Date.parse(normalizeDuplicateTime(raw));
return isNaN(t) ? -1 : t;
}
// 「查看明细」按钮点击开抽屉(动态元素,事件委托)
function handleDuplicateDetailClick(event) {
var button = event.target.closest('[data-open-asin-detail]');
if (button) openShopDataDuplicateDrawer(button.dataset.openAsinDetail);
}
// 撞款详情卡片区:同一 ASIN 在多条店铺的上架明细(仅跨店铺重复,按店铺数倒序)
function renderShopDataDuplicateDetailCards() {
var block = document.getElementById('dupCheckDetailBlock');
var cardsEl = document.getElementById('dupCheckDetailCards');
if (!block || !cardsEl) return;
var repeated = (shopDataDuplicateItems || []).filter(function (item) {
return Number(item.shop_count) >= 2;
}).slice().sort(function (a, b) {
return (Number(b.shop_count) - Number(a.shop_count)) || dupTimeRank(b.first_date || '') - dupTimeRank(a.first_date || '');
});
if (!repeated.length) {
block.style.display = 'none';
cardsEl.innerHTML = '';
return;
}
var header = [
{ label: 'ASIN', width: '14%' },
{ label: '店铺数', width: '8%' },
{ label: '记录条数', width: '10%' },
{ label: '品牌', width: '12%' },
{ label: '对应店铺', width: '22%' },
{ label: '国家', width: '14%' },
{ label: '上架时间', width: '12%' },
{ label: '操作', width: '8%' }
];
var tbody = shopDataDuplicateItems.map(function (item) {
return renderShopDataDuplicateRow(item);
}).join('');
list.innerHTML = '<table>' + shopDataDuplicateHeader(header) + '<tbody>' + tbody + '</tbody></table>';
}
function renderShopDataDuplicateRow(item) {
var occurrences = Array.isArray(item.occurrences) ? item.occurrences : [];
var brand = '';
var countries = {};
var shops = {};
occurrences.forEach(function (occ) {
if (occ.brand && !brand) brand = occ.brand;
(occ.country_codes || []).forEach(function (c) { countries[c] = true; });
shops[occ.shop_name || '-'] = true;
block.style.display = '';
cardsEl.innerHTML = repeated.map(function (item) {
// 按店铺分组:名称 + 次数 + 站点 + 上架时间(倒序)
var shopsMap = {};
(item.occurrences || []).forEach(function (occ) {
var key = occ.shop_name || '-';
if (!shopsMap[key]) shopsMap[key] = { shop: key, sites: {}, times: [], count: 0 };
var info = shopsMap[key];
info.count++;
if (occ.country) info.sites[occ.country] = true;
var d = normalizeDuplicateTime(occ.date);
if (d) info.times.push(d);
});
var countryLabel = countryListLabel(Object.keys(countries));
var shopList = Object.keys(shops).join('、');
var dateList = occurrences.map(function (occ) { return occ.date; })
.filter(function (d) { return d; }).join('、');
return '<tr data-duplicate-asin="' + escapeHtml(item.asin) + '">' +
'<td><span class="dup-asin">' + escapeHtml(item.asin) + '</span></td>' +
'<td><span class="dup-count">' + (item.shop_count || 1) + ' 家</span></td>' +
'<td>' + (item.record_count || occurrences.length) + ' 条</td>' +
'<td>' + escapeHtml(brand || '-') + '</td>' +
'<td>' + escapeHtml(shopList || '-') + '</td>' +
'<td>' + escapeHtml(countryLabel || '-') + '</td>' +
'<td>' + escapeHtml(dateList || '-') + '</td>' +
'<td><button class="btn btn-sm" type="button" data-dup-toggle>详情</button></td>' +
'</tr>' + renderShopDataDuplicateDetailRows(item);
}
function renderShopDataDuplicateDetailRows(item) {
var occurrences = Array.isArray(item.occurrences) ? item.occurrences : [];
if (!occurrences.length) return '';
var rows = occurrences.map(function (occ) {
var countries = countryListLabel(occ.country_codes);
return '<tr class="dup-detail-row" style="display:none;">' +
'<td></td>' +
'<td class="dup-shop">' + escapeHtml(occ.shop_name || '-') + '</td>' +
'<td class="dup-country">' + escapeHtml(occ.group_name || '-') + '</td>' +
'<td class="dup-country">' + escapeHtml(countries) + '</td>' +
'<td class="dup-date">' + escapeHtml(occ.date || '-') + '</td>' +
'<td class="dup-date">' + escapeHtml(occ.price || '-') + '</td>' +
'<td>' + escapeHtml(occ.brand || '-') + '</td>' +
'<td></td>' +
var rows = Object.keys(shopsMap).sort().map(function (key) {
var info = shopsMap[key];
info.times.sort(function (a, b) { return (a < b ? 1 : (a > b ? -1 : 0)); });
info.times = Array.from(new Set(info.times));
var rowSites = Object.keys(info.sites).sort().map(function (site) {
return '<span class="dup-check-site">' + escapeHtml(site) + '</span>';
}).join('');
return '<tr>' +
'<td class="dup-shop">' + escapeHtml(info.shop) + '<span class="dup-detail-count">' + info.count + ' 次</span></td>' +
'<td>' + (rowSites || '-') + '</td>' +
'<td class="dup-date">' + info.times.map(escapeHtml).join('') + '</td>' +
'</tr>';
}).join('');
return rows;
return '<div class="duplicate-asin-card">' +
'<div class="duplicate-asin-card-head">' +
'<span class="dup-asin" title="' + escapeHtml(item.asin) + '">' + escapeHtml(item.asin) + '</span>' +
'<span class="dup-count">' + item.shop_count + ' 家店铺</span>' +
'<span class="dup-brand">' + escapeHtml(item.brand || '') + '</span>' +
'<button class="btn btn-sm btn-secondary dup-detail-view-btn" type="button" data-open-asin-detail="' + escapeHtml(item.asin) + '">查看明细</button>' +
'</div>' +
'<table class="duplicate-asin-table dup-card-table"><thead>' +
'<tr><th>店铺</th><th>站点</th><th>上架时间</th></tr></thead>' +
'<tbody>' + rows + '</tbody></table>' +
'</div>';
}).join('');
}
function loadShopDataDuplicateAsins(page) {
function openShopDataDuplicateDrawer(asin) {
var item = null;
(shopDataDuplicateItems || []).forEach(function (it) { if (it.asin === asin) item = it; });
if (!item) return;
var occurrences = item.occurrences || [];
var brand = '', dateMin = '', dateMax = '', sites = {}, prices = [];
var shopsMap = {};
occurrences.forEach(function (occ) {
if (!brand && occ.brand) brand = occ.brand;
var d = normalizeDuplicateTime(occ.date);
if (d) {
if (!dateMin || d < dateMin) dateMin = d;
if (!dateMax || d > dateMax) dateMax = d;
}
if (occ.country) sites[occ.country] = true;
if (occ.price && prices.indexOf(occ.price) < 0) prices.push(occ.price);
var shopKey = occ.shop_name || '-';
if (!shopsMap[shopKey]) {
shopsMap[shopKey] = { shop: occ.shop_name || '-', group: occ.group_name || '', sites: {}, times: {}, price: '', count: 0 };
}
var info = shopsMap[shopKey];
info.count++;
if (occ.country) info.sites[occ.country] = true;
if (d) info.times[d] = true;
if (occ.price && !info.price) info.price = occ.price;
});
var dateRange = dateMin && dateMax ? (dateMin === dateMax ? dateMin : dateMin + ' ~ ' + dateMax) : '-';
var priceRange = prices.length ? prices.slice(0, 4).join(' / ') + (prices.length > 4 ? ' 等' : '') : '-';
var siteBadges = Object.keys(sites).sort().map(function (site) {
return '<span class="dup-check-site">' + escapeHtml(site) + '</span>';
}).join('');
var rows = Object.keys(shopsMap).sort().map(function (key) {
var info = shopsMap[key];
var times = Object.keys(info.times).sort().reverse();
var rowSites = Object.keys(info.sites).sort().map(function (site) {
return '<span class="dup-check-site">' + escapeHtml(site) + '</span>';
}).join('');
return '<tr>' +
'<td><span class="dup-asin-cell">' + escapeHtml(info.shop) + '</span></td>' +
'<td>' + escapeHtml(info.group || '-') + '</td>' +
'<td>' + (rowSites || '-') + '</td>' +
'<td class="dup-date">' + times.map(escapeHtml).join('、') + '</td>' +
'<td class="dup-date">' + escapeHtml(info.price || '-') + '</td>' +
'<td>' + info.count + '</td></tr>';
}).join('');
document.getElementById('dupCheckDrawerAsin').textContent = item.asin;
document.getElementById('dupCheckDrawerSubtitle').textContent =
item.shop_count + ' 家店铺 · ' + item.record_count + ' 条上架记录';
document.getElementById('dupCheckDrawerBody').innerHTML =
'<dl class="dup-check-drawer-meta">' +
'<dt>品牌</dt><dd>' + escapeHtml(brand || '-') + '</dd>' +
'<dt>价格</dt><dd>' + escapeHtml(priceRange) + '</dd>' +
'<dt>日期范围</dt><dd>' + escapeHtml(dateRange) + '</dd>' +
'<dt>站点</dt><dd>' + (siteBadges || '-') + '</dd></dl>' +
'<h4 style="margin:14px 0 6px;font-size:13.5px;">店铺 / 站点 / 上架时间 / 次数</h4>' +
'<div class="table-scroll"><table class="dup-check-drawer-table">' +
'<thead><tr><th>店铺</th><th>分组</th><th>站点</th><th>上架时间</th><th>价格</th><th>次数</th></tr></thead>' +
'<tbody>' + rows + '</tbody></table></div>';
document.getElementById('dupCheckDrawerMask').classList.add('show');
}
function closeShopDataDuplicateDrawer() {
document.getElementById('dupCheckDrawerMask').classList.remove('show');
}
function loadShopDataDuplicateCheckOverview(force) {
var progress = document.getElementById('shopDataDuplicateProgress');
var button = document.getElementById('btnRefreshShopDataDuplicates');
if (button) button.disabled = true;
if (force) progress.textContent = '正在全量扫描各店铺结果文件,请稍候...';
fetch('/api/admin/shop-data-crawl/duplicate-check-overview' + (force ? '?force=1' : ''))
.then(function (response) { return response.json(); })
.then(function (res) {
if (!res.success) throw new Error(res.error || '加载失败');
shopDataDuplicateOverviewPending = !!res.pending;
shopDataDuplicateScannedAt = res.scanned_at || '';
var summary = res.summary || {};
renderShopDataDuplicateMetrics(summary);
renderShopDataDuplicateDistribution(res.shops || []);
var dupTabCount = document.getElementById('dupCheckMonitorCount');
var allTabCount = document.getElementById('dupCheckAllCount');
if (dupTabCount) dupTabCount.textContent = summary.duplicate_asin_total != null ? summary.duplicate_asin_total : 0;
if (allTabCount) allTabCount.textContent = summary.asin_total != null ? summary.asin_total : 0;
loadShopDataDuplicateCheckItems(1);
})
.catch(function (error) {
var metricsEl = document.getElementById('shopDataDuplicateMetrics');
if (metricsEl) {
metricsEl.innerHTML = '<div class="shop-data-empty-hint">加载失败:'
+ escapeHtml(error.message || '') + '</div>';
}
})
.finally(function () {
if (button) button.disabled = false;
if (progress) progress.textContent = '';
});
}
function loadShopDataDuplicateCheckItems(page) {
if (shopDataDuplicateLoading) return;
shopDataDuplicatePage = page || 1;
shopDataDuplicateLoading = true;
var list = document.getElementById('shopDataDuplicateList');
var progress = document.getElementById('shopDataDuplicateProgress');
var button = document.getElementById('btnRefreshShopDataDuplicates');
progress.textContent = '正在读取各店铺结果文件并分析,请稍候...';
button.disabled = true;
list.innerHTML = '<div class="shop-data-empty-hint">分析中...</div>';
fetch('/api/admin/shop-data-crawl/duplicate-asins?' + buildShopDataDuplicateQuery(shopDataDuplicatePage))
var exportButton = document.getElementById('btnExportShopDataDuplicates');
list.innerHTML = '<div class="shop-data-empty-hint">加载中...</div>';
if (exportButton) exportButton.disabled = true;
fetch('/api/admin/shop-data-crawl/duplicate-check-items?' + buildShopDataDuplicateQuery(shopDataDuplicatePage))
.then(function (response) { return response.json(); })
.then(function (res) {
if (!res.success) throw new Error(res.error || '分析失败');
if (!res.success) throw new Error(res.error || '加载失败');
shopDataDuplicateItems = res.items || [];
shopDataDuplicateShops = res.shops || [];
shopDataDuplicateTotal = Number(res.total) || 0;
shopDataDuplicateAnalyzed.shopCount = Number(res.analyzed_shop_count) || 0;
shopDataDuplicateAnalyzed.resultCount = Number(res.analyzed_result_count) || 0;
if (res.scanned_at) shopDataDuplicateScannedAt = res.scanned_at;
var totalEl = document.getElementById('shopDataDuplicateTotal');
totalEl.textContent = '共 ' + shopDataDuplicateTotal + ' 个重复 ASIN · 已分析 ' + shopDataDuplicateAnalyzed.shopCount + ' 家店铺';
var tabCount = document.getElementById('shopDataDuplicateTabCount');
if (tabCount) tabCount.textContent = shopDataDuplicateTotal;
renderShopDataDuplicateList();
renderPagination('shopDataDuplicatePagination', shopDataDuplicateTotal, shopDataDuplicatePage, shopDataDuplicatePageSize, loadShopDataDuplicateAsins);
totalEl.textContent = (shopDataDuplicateView === 'monitor' ? '重复监控' : '全部ASIN台账')
+ ' · 共 ' + shopDataDuplicateTotal + ' 个 ASIN'
+ (shopDataDuplicateScannedAt ? ' · 扫描时间 ' + shopDataDuplicateScannedAt
: (shopDataDuplicateOverviewPending ? ' · 尚无扫描结果(点击「重新分析」立即扫描)' : ''));
renderShopDataDuplicateMatrix();
renderShopDataDuplicateDetailCards();
renderPagination('shopDataDuplicatePagination', shopDataDuplicateTotal,
shopDataDuplicatePage, shopDataDuplicatePageSize, loadShopDataDuplicateCheckItems);
})
.catch(function (error) {
shopDataDuplicateItems = [];
shopDataDuplicateTotal = 0;
list.innerHTML = '<div class="shop-data-empty-hint">分析失败:' + escapeHtml(error.message || '') + '</div>';
list.innerHTML = '<div class="shop-data-empty-hint">加载失败:' + escapeHtml(error.message || '') + '</div>';
document.getElementById('shopDataDuplicateTotal').textContent = '';
renderShopDataDuplicateDetailCards();
})
.finally(function () {
shopDataDuplicateLoading = false;
progress.textContent = '';
button.disabled = false;
if (progress) progress.textContent = '';
if (exportButton) exportButton.disabled = false;
});
}
function switchShopDataSubTab(view) {
var recordsView = document.getElementById('shopDataRecordsView');
var duplicatesView = document.getElementById('shopDataDuplicatesView');
var recordsTab = document.getElementById('shopDataSubTabRecords');
var duplicatesTab = document.getElementById('shopDataSubTabDuplicates');
var recordsActive = view === 'records';
recordsView.style.display = recordsActive ? '' : 'none';
duplicatesView.style.display = recordsActive ? 'none' : '';
recordsTab.classList.toggle('active', recordsActive);
recordsTab.setAttribute('aria-selected', recordsActive ? 'true' : 'false');
duplicatesTab.classList.toggle('active', !recordsActive);
duplicatesTab.setAttribute('aria-selected', !recordsActive ? 'true' : 'false');
if (!recordsActive) {
loadShopDataDuplicateAsins(1);
function switchDupCheckView(view) {
if (view === shopDataDuplicateView) return;
shopDataDuplicateView = view;
var monitorTab = document.getElementById('dupCheckTabMonitor');
var allTab = document.getElementById('dupCheckTabAll');
var monitorActive = view === 'monitor';
monitorTab.classList.toggle('active', monitorActive);
monitorTab.setAttribute('aria-selected', monitorActive ? 'true' : 'false');
allTab.classList.toggle('active', !monitorActive);
allTab.setAttribute('aria-selected', !monitorActive ? 'true' : 'false');
loadShopDataDuplicateCheckItems(1);
}
function exportShopDataDuplicates() {
if (shopDataDuplicateExporting) return;
shopDataDuplicateExporting = true;
var progress = document.getElementById('shopDataDuplicateProgress');
var button = document.getElementById('btnExportShopDataDuplicates');
if (button) button.disabled = true;
if (progress) progress.textContent = '正在导出当前筛选结果...';
var query = new URLSearchParams(buildShopDataDuplicateQuery(1));
query.delete('page');
query.delete('page_size');
fetch('/api/admin/shop-data-crawl/duplicate-check-export?' + query.toString())
.then(function (response) {
if (!response.ok) {
return response.json().then(function (data) { throw new Error(data.error || '导出失败'); });
}
return response.blob();
})
.then(function (blob) {
var url = URL.createObjectURL(blob);
var a = document.createElement('a');
a.href = url;
var now = new Date();
function pad(n) { return n < 10 ? '0' + n : String(n); }
a.download = '店铺数据重复检查_' + now.getFullYear() + pad(now.getMonth() + 1) + pad(now.getDate())
+ '_' + pad(now.getHours()) + pad(now.getMinutes()) + '.csv';
document.body.appendChild(a);
a.click();
document.body.removeChild(a);
setTimeout(function () { URL.revokeObjectURL(url); }, 1000);
})
.catch(function (error) {
if (progress) progress.textContent = '导出失败:' + (error.message || '');
})
.finally(function () {
shopDataDuplicateExporting = false;
if (button) button.disabled = false;
if (progress) progress.textContent = '';
});
}
function downloadShopDataTask(item) {
@@ -2223,27 +2461,33 @@
document.getElementById('btnFilterShopDataTasks').onclick = function () {
loadShopDataCrawlTasks(1);
if (document.getElementById('shopDataDuplicatesView').style.display !== 'none') {
loadShopDataDuplicateAsins(1);
}
};
document.getElementById('btnResetShopDataTasks').onclick = function () {
['shopDataTaskFilterShop', 'shopDataTaskFilterGroup', 'shopDataTaskFilterCountry', 'shopDataTaskFilterFrom', 'shopDataTaskFilterTo']
.forEach(function (id) { document.getElementById(id).value = ''; });
loadShopDataCrawlTasks(1);
if (document.getElementById('shopDataDuplicatesView').style.display !== 'none') {
loadShopDataDuplicateAsins(1);
}
};
document.getElementById('shopDataSubTabRecords').onclick = function () { switchShopDataSubTab('records'); };
document.getElementById('shopDataSubTabDuplicates').onclick = function () { switchShopDataSubTab('duplicates'); };
document.getElementById('btnRefreshShopDataDuplicates').onclick = function () { loadShopDataDuplicateAsins(1); };
document.getElementById('btnFilterShopDataDuplicates').onclick = function () { loadShopDataDuplicateAsins(1); };
document.getElementById('btnRefreshShopDataDuplicates').onclick = function () { loadShopDataDuplicateCheckOverview(true); };
document.getElementById('btnFilterShopDataDuplicates').onclick = function () { loadShopDataDuplicateCheckItems(1); };
document.getElementById('btnResetShopDataDuplicates').onclick = function () {
['shopDataDupFilterAsin', 'shopDataDupFilterShop', 'shopDataDupFilterCountry', 'shopDataDupFilterDateFrom', 'shopDataDupFilterDateTo']
['shopDataDupFilterAsin', 'shopDataDupFilterShop', 'shopDataDupFilterCountry', 'shopDataDupFilterSite', 'shopDataDupFilterDateFrom', 'shopDataDupFilterDateTo']
.forEach(function (id) { document.getElementById(id).value = ''; });
loadShopDataDuplicateAsins(1);
loadShopDataDuplicateCheckItems(1);
};
document.getElementById('dupCheckTabMonitor').onclick = function () { switchDupCheckView('monitor'); };
document.getElementById('dupCheckTabAll').onclick = function () { switchDupCheckView('all'); };
document.getElementById('btnExportShopDataDuplicates').onclick = exportShopDataDuplicates;
document.getElementById('btnCloseDupCheckDrawer').onclick = closeShopDataDuplicateDrawer;
document.getElementById('dupCheckDrawerMask').onclick = function (event) {
if (event.target === this) closeShopDataDuplicateDrawer();
};
// 矩阵表格内 ASIN / 数字徽标点击打开抽屉(动态元素,事件委托)
document.getElementById('shopDataDuplicateList').onclick = function (event) {
var target = event.target.closest('[data-open-drawer]');
if (target) openShopDataDuplicateDrawer(target.dataset.openDrawer);
};
// 撞款详情卡片区「查看明细」按钮(动态元素,事件委托)
document.getElementById('dupCheckDetailCards').onclick = handleDuplicateDetailClick;
// 全选移到表格表头后为动态元素,用事件委托
document.getElementById('shopDataTaskGrid').onchange = function (event) {
var checkbox = event.target.closest('[data-shop-data-select]');
@@ -2253,9 +2497,10 @@
if (checkbox.checked) selectedShopDataResultIds.add(resultId);
else selectedShopDataResultIds.delete(resultId);
syncShopDataSelectionUi();
return;
}
if (event.target.closest('#shopDataTaskSelectAll')) {
};
// 全选移到工具栏后不在 grid 容器内,单独绑定 change 事件
document.getElementById('shopDataTaskSelectAll').onchange = function (event) {
shopDataTasks.forEach(function (item) {
var sid = shopDataResultId(item);
if (!item.file_ready || !sid) return;
@@ -2263,7 +2508,6 @@
else selectedShopDataResultIds.delete(sid);
});
syncShopDataSelectionUi();
}
};
document.getElementById('shopDataTaskGrid').onclick = function (event) {
var downloadButton = event.target.closest('[data-shop-data-download]');
@@ -2279,28 +2523,6 @@
}
};
document.getElementById('btnBatchDownloadShopDataTasks').onclick = downloadShopDataTasksZip;
document.getElementById('shopDataDuplicateList').onclick = function (event) {
var toggle = event.target.closest('[data-dup-toggle]');
if (!toggle) return;
var row = toggle.closest('tr');
var allDetailRows = document.querySelectorAll('.dup-detail-row');
// 若当前 ASIN 的详情行已展开,则本次点击收起
var cur = row ? row.nextElementSibling : null;
var wasExpanded = false;
while (cur && cur.classList && cur.classList.contains('dup-detail-row')) {
if (cur.style.display !== 'none') { wasExpanded = true; break; }
cur = cur.nextElementSibling;
}
allDetailRows.forEach(function (dr) { dr.style.display = 'none'; });
if (row && !wasExpanded) {
// 展开当前 ASIN 的详情行(紧随其后)
cur = row.nextElementSibling;
while (cur && cur.classList && cur.classList.contains('dup-detail-row')) {
cur.style.display = '';
cur = cur.nextElementSibling;
}
}
};
document.getElementById('btnOpenShopDataTaskPermissions').onclick = openShopDataTaskPermissions;
document.getElementById('btnCloseShopDataTaskPermissions').onclick = closeShopDataTaskPermissions;
document.getElementById('btnCancelShopDataTaskPermissions').onclick = closeShopDataTaskPermissions;
@@ -2541,30 +2763,160 @@
String(u.created_by_id || '') === String(leaderUserId || '');
});
}
// ========== 分组管理通讯录:按拼音分组 + 模糊搜索 + 字母索引 ==========
// 隐藏 selectshopManageGroupMemberSelect)仅作为数据交换层,保存逻辑不变。
var shopManageGroupEligibleUsers = [];
var shopManageGroupSelectedIds = {};
var shopManageGroupSearchKeyword = '';
var shopManageGroupContactGroupsCache = [];
var SHOP_GROUP_AVATAR_COLORS = ['#f97316', '#0ea5e9', '#8b5cf6', '#10b981', '#ef4444', '#eab308', '#14b8a6', '#6366f1'];
function shopGroupAvatarColor(name) {
var code = String(name || '#').charCodeAt(0) || 0;
return SHOP_GROUP_AVATAR_COLORS[code % SHOP_GROUP_AVATAR_COLORS.length];
}
function shopGroupUserInitial(user) {
var abbr = String(user.pinyin_abbr || '').trim();
if (abbr) {
var first = abbr.charAt(0).toUpperCase();
return /^[A-Z0-9]$/.test(first) ? first : '#';
}
var name = String(user.username || '').trim();
if (!name) return '#';
var first = name.charAt(0);
if (/[A-Za-z0-9]/.test(first)) return first.toUpperCase();
return '#';
}
function shopGroupMatchSearch(user, keyword) {
if (!keyword) return true;
var kw = keyword.toLowerCase();
if (String(user.username || '').toLowerCase().indexOf(kw) !== -1) return true;
if (String(user.pinyin_abbr || '').toLowerCase().indexOf(kw) !== -1) return true;
return false;
}
function buildShopManageGroupContactGroups(users) {
var groups = {};
users.forEach(function (u) {
var initial = shopGroupUserInitial(u);
var key = /^[A-Z]$/.test(initial) ? initial : '#';
(groups[key] = groups[key] || []).push(u);
});
var keys = Object.keys(groups).sort(function (a, b) {
if (a === '#') return 1;
if (b === '#') return -1;
return a < b ? -1 : 1;
});
return keys.map(function (key) {
var list = groups[key].slice().sort(function (a, b) {
return String(a.username || '').localeCompare(String(b.username || ''), 'zh');
});
return { key: key, items: list };
});
}
function renderShopManageGroupContact() {
var contactEl = document.getElementById('shopManageGroupContact');
var indexEl = document.getElementById('shopManageGroupMemberIndex');
if (!contactEl) return;
var keyword = (shopManageGroupSearchKeyword || '').trim();
var filtered = shopManageGroupEligibleUsers.filter(function (u) {
return shopGroupMatchSearch(u, keyword);
});
if (!shopManageGroupEligibleUsers.length) {
contactEl.innerHTML = '<div class="shop-group-contact-empty">当前组长名下暂无可添加的普通员工账号。</div>';
if (indexEl) indexEl.innerHTML = '';
return;
}
if (!filtered.length) {
contactEl.innerHTML = '<div class="shop-group-contact-empty">未找到与「' + escapeHtml(keyword) + '」匹配的用户,可尝试输入拼音首字母,如 zwh。</div>';
if (indexEl) indexEl.innerHTML = '';
return;
}
shopManageGroupContactGroupsCache = buildShopManageGroupContactGroups(filtered);
contactEl.innerHTML = shopManageGroupContactGroupsCache.map(function (group) {
var itemsHtml = group.items.map(function (u) {
var selected = !!shopManageGroupSelectedIds[String(u.id)];
var name = u.username || '';
return '<div class="shop-group-contact-item' + (selected ? ' is-selected' : '') + '" data-user-id="' + escapeHtml(u.id) + '" title="' + escapeHtml(name) + '">' +
'<span class="shop-group-item-avatar" style="background:' + shopGroupAvatarColor(name) + ';">' + escapeHtml(shopGroupUserInitial(u)) + '</span>' +
'<span class="shop-group-item-name">' + escapeHtml(name) + '</span>' +
'<span class="shop-group-item-check" aria-hidden="true"><svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-linecap="round" stroke-linejoin="round"><path d="M20 6 9 17l-5-5"></path></svg></span>' +
'</div>';
}).join('');
return '<div class="shop-group-contact-group" data-letter="' + escapeHtml(group.key) + '">' + escapeHtml(group.key) + '</div>' +
'<div class="shop-group-contact-items">' + itemsHtml + '</div>';
}).join('');
if (indexEl) {
indexEl.innerHTML = shopManageGroupContactGroupsCache.map(function (group) {
return '<button type="button" data-letter="' + escapeHtml(group.key) + '">' + escapeHtml(group.key) + '</button>';
}).join('');
}
}
function syncShopManageGroupMemberSelect() {
var sel = document.getElementById('shopManageGroupMemberSelect');
if (!sel) return;
sel.innerHTML = shopManageGroupEligibleUsers.map(function (u) {
var selected = !!shopManageGroupSelectedIds[String(u.id)];
return '<option value="' + escapeHtml(u.id) + '"' + (selected ? ' selected' : '') + '>' + escapeHtml(u.username || '') + '</option>';
}).join('');
}
function updateShopManageGroupMemberCount() {
var countEl = document.getElementById('shopManageGroupMemberCount');
if (countEl) {
countEl.textContent = '已选 ' + Object.keys(shopManageGroupSelectedIds).length + ' 人';
}
}
function toggleShopManageGroupMember(userId) {
var key = String(userId || '');
if (!key) return;
var contactEl = document.getElementById('shopManageGroupContact');
var itemEl = contactEl ? Array.prototype.find.call(
contactEl.querySelectorAll('.shop-group-contact-item'),
function (el) { return el.dataset.userId === key; }
) : null;
if (shopManageGroupSelectedIds[key]) {
delete shopManageGroupSelectedIds[key];
} else {
shopManageGroupSelectedIds[key] = true;
}
if (itemEl) itemEl.classList.toggle('is-selected', !!shopManageGroupSelectedIds[key]);
syncShopManageGroupMemberSelect();
updateShopManageGroupMemberCount();
}
function shopGroupScrollToLetter(letter) {
var contactEl = document.getElementById('shopManageGroupContact');
if (!contactEl) return;
var groupEl = Array.prototype.find.call(
contactEl.querySelectorAll('.shop-group-contact-group'),
function (el) { return el.dataset.letter === letter; }
);
if (groupEl) groupEl.scrollIntoView({ behavior: 'smooth', block: 'start' });
}
function resetShopManageGroupSearch() {
shopManageGroupSearchKeyword = '';
var searchInput = document.getElementById('shopManageGroupMemberSearch');
if (searchInput) searchInput.value = '';
}
function refreshShopManageGroupMemberSelect(leaderUserId, selectedUserIds) {
var sel = document.getElementById('shopManageGroupMemberSelect');
var helpEl = document.getElementById('shopManageGroupMemberHelp');
if (!sel) return;
var selectedMap = {};
resetShopManageGroupSearch();
shopManageGroupEligibleUsers = getEligibleShopManageGroupUsers(leaderUserId);
shopManageGroupSelectedIds = {};
(selectedUserIds || []).forEach(function (id) {
selectedMap[String(id)] = true;
shopManageGroupSelectedIds[String(id)] = true;
});
var eligibleUsers = getEligibleShopManageGroupUsers(leaderUserId);
var options = eligibleUsers.map(function (u) {
return '<option value="' + escapeHtml(u.id) + '"' + (selectedMap[String(u.id)] ? ' selected' : '') + '>' + escapeHtml(u.username || '') + '</option>';
});
if (options.length) {
sel.innerHTML = options.join('');
syncShopManageGroupMemberSelect();
renderShopManageGroupContact();
updateShopManageGroupMemberCount();
if (helpEl) {
helpEl.textContent = '可添加当前组长创建的普通员工账号,按住 Ctrl 或 Command 可多选。';
}
return;
}
sel.innerHTML = '<option value="" disabled>当前组长暂无可添加组员</option>';
if (helpEl) {
helpEl.textContent = currentUserRole === 'normal'
helpEl.textContent = shopManageGroupEligibleUsers.length
? '可添加当前组长创建的普通员工账号,点击用户即可选中 / 取消;顶部支持按用户名或拼音首字母搜索。'
: (currentUserRole === 'normal'
? '普通账号没有下属普通员工时,这里会为空;当前账号只能作为组长使用。'
: '当前组长名下暂无可添加的普通员工账号。';
: '当前组长名下暂无可添加的普通员工账号。');
}
}
function setShopManageGroupLeader(leaderUserId, leaderUsername, selectedUserIds) {
@@ -2595,7 +2947,8 @@
id: u.id,
username: u.username || '',
role: u.role || 'normal',
created_by_id: u.created_by_id || null
created_by_id: u.created_by_id || null,
pinyin_abbr: u.pinyin_abbr || ''
};
});
setShopManageGroupLeader(
@@ -3891,6 +4244,28 @@
document.getElementById('btnCloseShopManageGroupModal').onclick = function () {
document.getElementById('shopManageGroupModal').classList.remove('show');
};
// 组员通讯录:搜索 / 字母索引 / 点击选中(打开弹窗时 refresh 会重置搜索词)
var groupMemberSearchInput = document.getElementById('shopManageGroupMemberSearch');
if (groupMemberSearchInput) {
groupMemberSearchInput.addEventListener('input', function () {
shopManageGroupSearchKeyword = groupMemberSearchInput.value || '';
renderShopManageGroupContact();
});
}
var groupMemberIndexEl = document.getElementById('shopManageGroupMemberIndex');
if (groupMemberIndexEl) {
groupMemberIndexEl.addEventListener('click', function (e) {
var btn = e.target.closest('button[data-letter]');
if (btn) shopGroupScrollToLetter(btn.dataset.letter);
});
}
var groupMemberContactEl = document.getElementById('shopManageGroupContact');
if (groupMemberContactEl) {
groupMemberContactEl.addEventListener('click', function (e) {
var item = e.target.closest('.shop-group-contact-item');
if (item) toggleShopManageGroupMember(item.dataset.userId);
});
}
document.getElementById('btnSearchShopManage').onclick = function () {
loadShopManage(1);
};
@@ -30,6 +30,7 @@ def _make_workbook(rows_by_sheet):
class ShopDataDuplicateAsinTest(unittest.TestCase):
def setUp(self):
self.app = Flask(__name__)
self.app.config['SECRET_KEY'] = 'test-secret'
self.group_rows = [
{'shop_name': 'Shop A', 'latest_created_at': datetime(2026, 8, 31, 1, 15, 1)},
{'shop_name': 'Shop B', 'latest_created_at': datetime(2026, 8, 31, 1, 15, 2)},
@@ -133,23 +134,38 @@ class ShopDataDuplicateAsinTest(unittest.TestCase):
self.shop_files[shop_name].save(stream)
return stream.getvalue()
def _run_request(self, query=''):
cursor = self._FakeCursor(
self.group_rows,
[self._result_row(i + 1, row['shop_name'], self.shop_country[row['shop_name']])
for i, row in enumerate(self.group_rows)],
)
connection = self._FakeConnection(cursor)
def _build_cache(self):
"""构造与旧扫描一致的缓存:3 家店铺、跨店重复 B0ABC111。"""
return {
'scanned_at': '2026-09-03 03:10:00',
'summary': {'shop_count': 3, 'total': 1},
'items': [{
'asin': 'B0ABC111',
'shop_count': 2,
'record_count': 3,
'occurrences': [
{'asin': 'B0ABC111', 'date': '2026-08-30', 'price': 'GBP 9.99', 'brand': 'BrandA',
'shop_name': 'Shop A', 'group_name': 'Group-Shop A', 'country_codes': ['UK', 'DE']},
{'asin': 'B0ABC111', 'date': '2026-08-30', 'price': 'EUR 10.99', 'brand': 'BrandA',
'shop_name': 'Shop A', 'group_name': 'Group-Shop A', 'country_codes': ['DE']},
{'asin': 'B0ABC111', 'date': '2026-08-31', 'price': 'GBP 8.50', 'brand': 'BrandA',
'shop_name': 'Shop B', 'group_name': 'Group-Shop B', 'country_codes': ['UK']},
],
}],
}
def _run_request(self, query='', cache=None):
# 界面默认读缓存;force 走实时扫描(需 mock 扫描核心)
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-asins?' + query):
with patch.object(admin_api, 'get_db', return_value=connection), \
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, '_latest_duplicate_scan', return_value=cache), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=(None, None, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=(None, None, None)), \
patch.object(admin_api, '_shop_data_crawl_fetch_result_bytes',
side_effect=lambda row: self._make_workbook_bytes(row['shop_name'])):
return admin_api.shop_data_crawl_duplicate_asins.__wrapped__()
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=(None, None, None)):
return admin_api.shop_data_crawl_duplicate_asins()
def test_detects_duplicate_asins_across_shops(self):
response = self._run_request('page=1&page_size=10')
response = self._run_request('page=1&page_size=10', cache=self._build_cache())
self.assertEqual(response.status_code, 200)
body = response.get_json()
self.assertTrue(body['success'])
@@ -165,44 +181,103 @@ class ShopDataDuplicateAsinTest(unittest.TestCase):
shop_a = next(occ for occ in item['occurrences'] if occ['shop_name'] == 'Shop A')
self.assertIn('UK', shop_a['country_codes'])
self.assertEqual(shop_a['date'], '2026-08-30')
self.assertEqual(body['scanned_at'], '2026-09-03 03:10:00')
def test_filters_asin_and_date_range(self):
# asin 模糊:匹配 B0ABC222 的只有 Shop B 一家,不构成跨店重复 → total 0
response = self._run_request('page=1&page_size=10&asin=B0ABC222')
def test_no_cache_returns_pending_empty(self):
# 无缓存时(定时任务尚未执行):返回空 + pending 提示
response = self._run_request('page=1&page_size=10', cache=None)
body = response.get_json()
self.assertTrue(body['success'])
self.assertEqual(body['total'], 0)
# 日期范围 08-31B0ABC111 在 Shop B 有该日记录 → 命中,展示完整记录(跨店仍 2 家)
response = self._run_request('page=1&page_size=10&date_from=2026-08-31&date_to=2026-08-31')
body = response.get_json()
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'][0]['asin'], 'B0ABC111')
self.assertEqual(body['items'][0]['shop_count'], 2) # 完整记录仍跨店
# 日期范围 08-30:Shop A 两条命中,同样返回完整记录
response = self._run_request('page=1&page_size=10&date_from=2026-08-30T00:00&date_to=2026-08-30T23:59')
body = response.get_json()
self.assertEqual(body['total'], 1)
def test_filters_shop_and_country(self):
# 店铺过滤 Shop B:命中含 Shop B 记录的 ASIN → B0ABC111,展示完整记录跨店
response = self._run_request('page=1&page_size=10&shop_name=Shop+B')
body = response.get_json()
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'][0]['asin'], 'B0ABC111')
# 国家过滤 FR:只有 Shop C 有 FR,其 ASIN 仅单店 → 不构成重复 → total 0
response = self._run_request('page=1&page_size=10&country=FR')
body = response.get_json()
self.assertEqual(body['total'], 0)
# 国家过滤 DEShop A 命中 → B0ABC111 完整记录 → total 1
response = self._run_request('page=1&page_size=10&country=DE')
body = response.get_json()
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'][0]['asin'], 'B0ABC111')
def test_pagination_when_page_out_of_range(self):
response = self._run_request('page=2&page_size=10')
body = response.get_json()
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'], [])
self.assertTrue(body['pending'])
def test_cache_filters_match_live_semantics(self):
cache = self._build_cache()
# 店铺过滤 Shop B:命中 → 完整记录
response = self._run_request('page=1&page_size=10&shop_name=Shop+B', cache=cache)
body = response.get_json()
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'][0]['asin'], 'B0ABC111')
# 日期范围 08-31 命中
response = self._run_request('page=1&page_size=10&date_from=2026-08-31&date_to=2026-08-31', cache=cache)
body = response.get_json()
self.assertEqual(body['total'], 1)
# 国家 FR 无命中
response = self._run_request('page=1&page_size=10&country=FR', cache=cache)
body = response.get_json()
self.assertEqual(body['total'], 0)
def test_force_runs_live_scan(self):
# force=1 触发实时扫描并保存缓存后返回
cursor = self._FakeCursor(
self.group_rows,
[self._result_row(i + 1, row['shop_name'], self.shop_country[row['shop_name']])
for i, row in enumerate(self.group_rows)],
)
connection = self._FakeConnection(cursor)
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-asins?page=1&page_size=10&force=1'):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, 'get_db', return_value=connection), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=(None, None, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=(None, None, None)), \
patch.object(admin_api, '_save_duplicate_scan', return_value=1), \
patch.object(admin_api, '_latest_duplicate_scan',
return_value=self._build_cache()), \
patch.object(admin_api, '_shop_data_crawl_fetch_result_bytes',
side_effect=lambda row: self._make_workbook_bytes(row['shop_name'])):
response = admin_api.shop_data_crawl_duplicate_asins()
self.assertIsNotNone(response)
body = response.get_json()
self.assertTrue(body['success'])
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'][0]['asin'], 'B0ABC111')
def test_force_conflict_when_lock_held(self):
# 锁被占(定时任务/他请求在扫)时 force 返回 409
cursor = self._FakeCursor(self.group_rows[:0], [])
connection = self._FakeConnection(cursor)
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-asins?page=1&page_size=10&force=1'):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, 'get_db', return_value=connection), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=(None, None, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=(None, None, None)), \
patch.object(admin_api, '_duplicate_scan_lock'):
admin_api._duplicate_scan_lock.acquire.return_value = False
response = admin_api.shop_data_crawl_duplicate_asins()
# 409 返回 (jsonify, status) tuple
self.assertEqual(response[1], 409)
def test_latest_duplicate_scan_deserializes_json_columns(self):
"""真实读库:summary_json / payload_json 为 JSON 字符串,需反序列化为 dict/list。"""
class _ScanCursor(self._FakeCursor):
def __init__(self):
self.kind = None
def execute(self, sql, params=()):
self.kind = 'scan'
def fetchone(self):
return {
'id': 9,
'summary_json': '{"shop_count": 3, "total": 1, "source": "import"}',
'payload_json': '[{"asin": "B0ABC111", "shop_count": 2}]',
'created_at': datetime(2026, 9, 3, 13, 35, 25),
}
class _ScanConnection(self._FakeConnection):
def cursor(self):
return _ScanCursor()
with patch.object(admin_api, 'get_db', return_value=_ScanConnection(None)):
cache = admin_api._latest_duplicate_scan()
self.assertIsNotNone(cache)
self.assertEqual(cache['summary']['shop_count'], 3)
self.assertEqual(cache['summary']['source'], 'import')
self.assertEqual(cache['items'][0]['asin'], 'B0ABC111')
self.assertEqual(cache['scanned_at'], '2026-09-03 13:35:25')
def test_parse_workbook_skips_unknown_sheets(self):
wb = _make_workbook({'英国': [('2026-08-30', 'B0TEST01', 'GBP 1.00', '')]})
@@ -0,0 +1,225 @@
"""店铺数据重复检查接口单元测试:全量缓存格式、矩阵分页、主管/超管数据范围过滤。"""
import sys
import unittest
from pathlib import Path
from unittest.mock import Mock, patch
from flask import Flask
sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
from blueprints import admin_api
def _shop(name, group, country_codes=None):
return {
'shop_name': name,
'group_name': group,
'country_codes': country_codes or [],
'rows': [],
}
class DuplicateCheckApiTest(unittest.TestCase):
"""直接对接口函数做单元测试:缓存通过 _latest_duplicate_scan mock 注入。"""
def setUp(self):
self.app = Flask(__name__)
self.app.config['SECRET_KEY'] = 'test-secret'
# 4 家店 / 3 个组;ASIN 分布:
# A1: ShopA(UK) + ShopB(UK) —— 跨店重复
# B1: ShopA(DE) 唯一
# C1: ShopC(FR) 唯一
# D1: ShopD(UK) 唯一
self.cache = {
'scanned_at': '2026-09-04 03:10:00',
'summary': {
'shop_count': 4, 'asin_total': 4, 'record_total': 5,
'duplicate_asin_total': 1, 'duplicate_shop_count': 2,
'site_count': 3, 'asin_per_shop': 1.2, 'source': 'job',
},
'shops': [
{'shop_name': 'ShopA', 'group_name': 'GroupA', 'country_codes': ['UK', 'DE'], 'asin_count': 2, 'record_count': 3},
{'shop_name': 'ShopB', 'group_name': 'GroupA', 'country_codes': ['UK'], 'asin_count': 1, 'record_count': 1},
{'shop_name': 'ShopC', 'group_name': 'GroupC', 'country_codes': ['FR'], 'asin_count': 1, 'record_count': 1},
{'shop_name': 'ShopD', 'group_name': 'GroupD', 'country_codes': ['UK'], 'asin_count': 1, 'record_count': 1},
],
'items': [
{'asin': 'A0000001', 'shop_count': 2, 'record_count': 2, 'occurrences': [
{'asin': 'A0000001', 'date': '2026-08-30', 'price': 'GBP 9.99', 'brand': 'BrandA',
'shop_name': 'ShopA', 'group_name': 'GroupA', 'country_codes': ['UK', 'DE'], 'country': 'UK'},
{'asin': 'A0000001', 'date': '2026-08-31', 'price': 'GBP 8.50', 'brand': 'BrandA',
'shop_name': 'ShopB', 'group_name': 'GroupA', 'country_codes': ['UK'], 'country': 'UK'},
]},
{'asin': 'B0000001', 'shop_count': 1, 'record_count': 1, 'occurrences': [
{'asin': 'B0000001', 'date': '2026-08-30', 'price': 'EUR 10.99', 'brand': 'BrandA',
'shop_name': 'ShopA', 'group_name': 'GroupA', 'country_codes': ['DE'], 'country': 'DE'},
]},
{'asin': 'C0000001', 'shop_count': 1, 'record_count': 1, 'occurrences': [
{'asin': 'C0000001', 'date': '2026-08-29', 'price': 'EUR 7.50', 'brand': 'BrandC',
'shop_name': 'ShopC', 'group_name': 'GroupC', 'country_codes': ['FR'], 'country': 'FR'},
]},
{'asin': 'D0000001', 'shop_count': 1, 'record_count': 1, 'occurrences': [
{'asin': 'D0000001', 'date': '2026-08-28', 'price': 'GBP 5.00', 'brand': 'BrandD',
'shop_name': 'ShopD', 'group_name': 'GroupD', 'country_codes': ['UK'], 'country': 'UK'},
]},
],
}
def _access_patches(self, role='super_admin', current_row=None):
return [
patch.object(admin_api, '_ensure_backend_menu_access',
return_value=(role, current_row or {'id': 1}, None)),
patch.object(admin_api, '_ensure_shop_data_crawl_data_access',
return_value=(role, current_row or {'id': 1}, None)),
patch.object(admin_api, '_latest_duplicate_scan', return_value=self.cache),
]
def _call(self, url, role='super_admin', current_row=None):
with self.app.test_request_context(url):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True):
for p in self._access_patches(role, current_row):
p.start()
try:
return admin_api.shop_data_crawl_duplicate_check_items()
finally:
for p in reversed(self._access_patches(role, current_row)):
p.stop()
def test_overview_super_admin_sees_all(self):
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-check-overview'):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=('super_admin', {'id': 1}, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=('super_admin', {'id': 1}, None)), \
patch.object(admin_api, '_latest_duplicate_scan', return_value=self.cache):
response = admin_api.shop_data_crawl_duplicate_check_overview()
body = response.get_json()
self.assertTrue(body['success'])
self.assertFalse(body['pending'])
self.assertEqual(body['summary']['asin_total'], 4)
self.assertEqual(body['summary']['record_total'], 5)
self.assertEqual(body['summary']['duplicate_asin_total'], 1)
self.assertEqual(len(body['shops']), 4)
def test_items_matix_columns_are_shops(self):
response = self._call('/api/admin/shop-data-crawl/duplicate-check-items?page=1&page_size=20&view=monitor')
body = response.get_json()
self.assertTrue(body['success'])
self.assertEqual(body['total'], 1) # monitor 只保留跨店重复
self.assertEqual(body['items'][0]['asin'], 'A0000001')
self.assertEqual([shop['shop_name'] for shop in body['shops']],
['ShopA', 'ShopB', 'ShopC', 'ShopD'])
def test_items_all_view_includes_unique_asins(self):
response = self._call('/api/admin/shop-data-crawl/duplicate-check-items?page=1&page_size=20&view=all')
body = response.get_json()
self.assertEqual(body['total'], 4)
def test_items_filter_by_asin_and_site(self):
response = self._call('/api/admin/shop-data-crawl/duplicate-check-items?page=1&page_size=20&asin=A0000001&site=UK')
body = response.get_json()
self.assertEqual(body['total'], 1)
self.assertEqual(body['items'][0]['asin'], 'A0000001')
def test_leader_sees_only_own_group_shops(self):
# 主管 id=653 只管理 GroupA(含 ShopA/ShopB
with patch.object(admin_api, '_shop_data_managed_shop_names',
return_value={'shopa', 'shopb'}):
response = self._call('/api/admin/shop-data-crawl/duplicate-check-items?page=1&page_size=20&view=all',
role='admin', current_row={'id': 653})
body = response.get_json()
self.assertEqual([shop['shop_name'] for shop in body['shops']], ['ShopA', 'ShopB'])
self.assertEqual(body['total'], 2) # A0000001(跨店)+ B0000001(唯一)
# 跨店 ASIN 在主管范围内仍是 2 家店
item = next(i for i in body['items'] if i['asin'] == 'A0000001')
self.assertEqual(item['shop_count'], 2)
def test_leader_overview_stats_recomputed_after_filter(self):
# 主管见 2 家店:唯一ASIN 2、上架记录 3、重复ASIN 1、重复店铺 2
with patch.object(admin_api, '_shop_data_managed_shop_names',
return_value={'shopa', 'shopb'}):
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-check-overview'):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=('admin', {'id': 653}, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=('admin', {'id': 653}, None)), \
patch.object(admin_api, '_latest_duplicate_scan', return_value=self.cache):
response = admin_api.shop_data_crawl_duplicate_check_overview()
body = response.get_json()
self.assertEqual(body['summary']['shop_count'], 2)
self.assertEqual(body['summary']['asin_total'], 2)
self.assertEqual(body['summary']['record_total'], 3)
self.assertEqual(body['summary']['duplicate_asin_total'], 1)
def test_leader_sees_nothing_when_no_group(self):
with patch.object(admin_api, '_shop_data_managed_shop_names', return_value=set()):
response = self._call('/api/admin/shop-data-crawl/duplicate-check-items?page=1&page_size=20&view=all',
role='admin', current_row={'id': 999999})
body = response.get_json()
self.assertEqual(body['shops'], [])
self.assertEqual(body['total'], 0)
def test_export_csv_contains_filtered_rows(self):
"""导出 CSV:行=上架记录,含 BOM,按筛选裁剪。"""
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-check-export?'):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=('super_admin', {'id': 1}, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=('super_admin', {'id': 1}, None)), \
patch.object(admin_api, '_latest_duplicate_scan', return_value=self.cache):
response = admin_api.shop_data_crawl_duplicate_check_export()
self.assertEqual(response.status_code, 200)
text = response.get_data(as_text=True)
self.assertTrue(text.startswith(''))
lines = text.lstrip('').strip().splitlines()
self.assertEqual(lines[0], 'ASIN,店铺数,店铺,分组,站点,上架时间,价格,品牌')
self.assertEqual(len(lines), 3) # header + 两条上架记录(A0000001 两店各一条)
def test_export_monitor_view_excludes_unique_asins(self):
with self.app.test_request_context('/api/admin/shop-data-crawl/duplicate-check-export?view=monitor'):
with patch('utils.auth.session', {'user_id': 1}), \
patch('utils.auth.is_session_user_valid', return_value=True), \
patch.object(admin_api, '_ensure_backend_menu_access', return_value=('super_admin', {'id': 1}, None)), \
patch.object(admin_api, '_ensure_shop_data_crawl_data_access', return_value=('super_admin', {'id': 1}, None)), \
patch.object(admin_api, '_latest_duplicate_scan', return_value=self.cache):
response = admin_api.shop_data_crawl_duplicate_check_export()
lines = response.get_data(as_text=True).lstrip('').strip().splitlines()
self.assertEqual(len(lines), 3)
self.assertIn('A0000001', lines[1])
def test_internal_request_falls_back_to_system_operator(self):
"""无请求上下文(定时扫描线程)时,内部请求用系统级超管作为 operatorId。"""
with patch('utils.auth.session', {'user_id': 1}): # 仅用于模拟无异常环境
# 无请求上下文:has_request_context() 为 False,直接走 _resolve_system_operator_id
with patch.object(admin_api, '_resolve_internal_token', return_value='test-token'), \
patch.object(admin_api, '_resolve_system_operator_id', return_value=7):
headers, params = admin_api._backend_java_internal_request()
self.assertEqual(headers.get('X-Internal-Token'), 'test-token')
self.assertEqual(params, {'operatorId': 7})
def test_internal_request_system_operator_missing_fails(self):
"""无请求上下文且系统中没有任何管理员时,直接报错而不是传 0。"""
with patch.object(admin_api, '_resolve_internal_token', return_value='test-token'), \
patch.object(admin_api, '_resolve_system_operator_id', return_value=None):
with self.assertRaises(ValueError):
admin_api._backend_java_internal_request()
def test_fetch_result_bytes_rejects_json_error_body(self):
"""Java 内部端点返回 JSON 错误体(如鉴权失败)时,抛出业务错误而不是 BadZipFile。"""
fake_response = Mock()
fake_response.raise_for_status = Mock()
fake_response.iter_content = Mock(return_value=[
'{"success":false,"message":"用户不存在","data":null,"code":401}'.encode('utf-8')])
fake_response.close = Mock()
with patch.object(admin_api, '_backend_java_internal_request',
return_value=({'X-Internal-Token': 'test-token'}, {'operatorId': 7})), \
patch.object(admin_api, '_get_backend_java_session') as fake_session:
fake_session.return_value.get = Mock(return_value=fake_response)
with self.assertRaises(ValueError) as ctx:
admin_api._shop_data_crawl_fetch_result_bytes({'result_id': 123})
self.assertIn('用户不存在', str(ctx.exception))
if __name__ == '__main__':
unittest.main()
File diff suppressed because it is too large Load Diff