收集更多信息

jetblue.py

 import platform
 import random
+import re
 import subprocess
 import time
 
@@ -23,7 +24,6 @@ def md5(text):
 # 连接MongoDB数据库
 client = MongoClient('mongodb://lcc-spider:JIUsfhiad&^@18.144.59.80:27017/?authSource=admin&directConnection=true')
 db = client.env
-
 # 查询所有文档
 # result = db.env_shape.find({'env.funcs' :{'$ne': None}})
 # data = {str(x['_id']): x['env'] for x in result}
@@ -85,6 +85,11 @@ for i in range(100):
             lastHash = m5
             usedTime = 0
             logger.debug('js变化')
+            token = re.findall('nt"\),\["(.+?)",', c)[0]
+            db.token_shape.insert_one({
+                '_id': token,
+                'update_time': int(time.time() * 1000)
+            })
             code = open('src.js', 'r', encoding='utf-8').read()
             code1 = open('js_dom_run.js', 'r', encoding='utf-8').read()
             with open('src_run1.js', 'w', encoding='utf-8') as f:
@@ -97,7 +102,9 @@ for i in range(100):
         # p = subprocess.Popen(['node', 'js_dom_run.js', selectId, '11'], stdin=subprocess.PIPE, stdout=subprocess.PIPE,
         # p = subprocess.Popen(['node', 'src_run_chrome_105.js', selectId, '11'], stdin=subprocess.PIPE, stdout=subprocess.PIPE,
         # p = subprocess.Popen(['node', 'src_run_company_chrome.js', selectId, '11'], stdin=subprocess.PIPE, stdout=subprocess.PIPE,
-        p = subprocess.Popen(['node', 'src_run1.js', selectId, '11'], stdin=subprocess.PIPE, stdout=subprocess.PIPE,
+        token = list(db.token_shape.aggregate( [ { '$sample': { 'size': 1 } } ] ))[0]['_id']
+
+        p = subprocess.Popen(['node', 'src_run1.js', selectId, '11', token], stdin=subprocess.PIPE, stdout=subprocess.PIPE,
                              stderr=subprocess.PIPE)
         output, err = p.communicate(timeout=10)
         tmp = ''