[+] Crawl index data

.gitignore

@@ -152,3 +152,4 @@ cython_debug/
 .idea/
 crawler
 .config
+index-data/

index_crawler.py

@@ -0,0 +1,73 @@
+from __future__ import annotations
+
+import datetime
+import json
+from pathlib import Path
+
+import requests
+
+
+ses = requests.Session()
+ses.headers = {'accept-language': 'zh-CN'}
+
+
+out_path = Path('index-data')
+out_path.mkdir(exist_ok=True)
+
+
+def write_entry(data):
+    id = data['guid']
+    time = datetime.datetime.fromtimestamp(data['delivery_time']).strftime('%Y-%m-%d %H-%M')
+
+    p = out_path / f'{time} {id}.json'
+    if p.is_file():
+        return False
+
+    p.write_text(json.dumps(data, ensure_ascii=False, indent=1))
+    return True
+
+
+def setup_proxy():
+    proxies = {
+        'http': 'socks5://localhost:9050',
+        'https': 'socks5://localhost:9050'
+    }
+
+    url = 'http://ifconfig.me/ip'
+
+    ip = requests.get(url).text.strip()
+    print(f'Raw ip: {ip}')
+
+    proxy_ip = requests.get(url, proxies=proxies).text.strip()
+    print(f'Proxy ip: {proxy_ip}')
+
+    assert ip != proxy_ip, 'Proxy did not start correctly.'
+
+
+if __name__ == '__main__':
+    setup_proxy()
+
+    prev_date_file = out_path / '0-prev-date.txt'
+
+    def send_req(prev_date: int | None):
+        add_param = {'logisticMinDeliveryTime': prev_date} if prev_date is not None else {}
+
+        r = requests.get('https://front.superbuy.com/logistic/get-index-pull-data',
+                         params={'onlyPackage': 1, **add_param}).json()
+
+        assert r['state'] == 0, 'Request failed.'
+
+        data = r['data']
+        successes = [write_entry(i) for i in data]
+
+        print(f'Out of {len(data)} entries, successfully wrote {sum(successes)} entries.')
+        prev_date_file.write_text(str(data[-1]['delivery_time'] - 1))
+
+    if not prev_date_file.is_file():
+        print('Getting first request...')
+        send_req(None)
+
+    while True:
+        prev_date = int(prev_date_file.read_text())
+        print(f'Getting entries before {prev_date}')
+        send_req(prev_date)

mobile.py

@@ -104,6 +104,9 @@ def create_diy_order(create: list[TaobaoOrder]):
 
     :param create: List of taobao urls and details
     """
+    orders = gateway_order_list()
+    ids = [get_url_param(i.GoodsLink, 'id') for o in orders for i in o.Items if 'id=' in i.GoodsLink]
+    create = [o for o in create if o.date >= '2022-08-10' and not any(get_url_param(i.url, 'id') in ids for i in o.items)]
     for c in create:
         shop_name = c.store.name
         shop_id = crawl(c.items[0].url)['data']['shop']['shopId']
@@ -127,7 +130,6 @@ def create_diy_order(create: list[TaobaoOrder]):
         resp = r.post('https://front.superbuy.com/order/transport/create-diy-order', data=j.encode('utf-8'),
                       headers={'content-type': 'application/json; charset=UTF-8'}).json()
         print(resp)
-        time.sleep(10)
 
 
 def fill_express_no(taobao_data: list[TaobaoOrder]):
@@ -177,9 +179,7 @@ app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_credentials=True)
 @app.post('/taobao/fill_items')
 def taobao_fill_items(body: Any = Body):
     taobao_data: list[TaobaoOrder] = js(body)
-    orders = gateway_order_list()
-    ids = [get_url_param(i.GoodsLink, 'id') for o in orders for i in o.Items if 'id=' in i.GoodsLink]
-    create_diy_order([o for o in taobao_data if o.date >= '2022-08-10' and not any(get_url_param(i.url, 'id') in ids for i in o.items)])
+    create_diy_order(taobao_data)
 
 
 @app.post('/taobao/fill_delivery')
@@ -190,6 +190,7 @@ def taobao_fill_delivery(body: Any = Body):
 
 
 if __name__ == '__main__':
-    print(login_cached(os.environ['user'], os.environ['pass']))
+    print(login(os.environ['user'], os.environ['pass']))
     # print(r.get(f'https://api.superbuy.com/gateway/oauth2/personalcenter/{USERID}').json())
+    create_diy_order(load_taobao())
     fill_express_no(load_taobao())