add mmic

plat/mmic/__init__.py

+#!/usr/bin/env python
+# -*- coding:utf-8 -*-
\ No newline at end of file

plat/mmic/cate_stock.py

+import os
+
+from utils.base import Module_Base
+domain='http://www.c-c.com'
+headers={
+"User-Agent":"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/72.0.3626.121 Safari/537.36",
+}
+a = Module_Base()
+# url = 'http://www.mmic.net.cn/'
+# resp = a.requests.get(url,headers=headers).text
+# dom = a.etree.HTML(resp)
+from lxml import etree
+print(os.getcwd())
+with open('plat/mmic/mmic.html',encoding='utf-8') as f:
+    text = f.read()
+dom_content=etree.HTML(text)
+level1_doms1=dom_content.xpath('//h2')
+level1_doms2=dom_content.xpath('//ul')
+print(len(level1_doms1))
+print(len(level1_doms2))
+level1_doms = zip(level1_doms1,level1_doms2)
+
+def get_levels():
+    levels = []
+    for level1_dom in level1_doms:
+        level1=dict()
+        level1['cat_name'] = level1_dom[0].xpath('.//a/text()')[0]
+        level1['url'] = level1_dom[0].xpath('.//a/@href')[0]
+        level1['level'] = 1
+        level1['islast'] = 0 if level1_dom[1].xpath('.//li/a') else 1
+        level1['parent_id'] = 0
+        level1['level2_list'] = []
+        print('一级分类',level1)
+        for level2_dom in level1_dom[1].xpath('.//li/a'):
+            level2 = {}
+            level2['cat_name'] = level2_dom.xpath('./text()')[0]
+            level2['url'] = domain + level2_dom.xpath('./@href')[0]
+            level2['level'] = 2
+            level2['islast'] = 1
+            level1['level2_list'].append(level2)
+        levels.append(level1)
+    return levels
+
+levels = get_levels()
+print(levels)
\ No newline at end of file

utils/mysqlopera.py

@@ -7,7 +7,7 @@ import pymysql
 
 
 COMPANY_PN_LIST = ['hc360', 'huangye88', 'pangwo', 'coovee', 'liebiao', 'china', 'ic114', 'cecb2b', 'trustexporter',
 COMPANY_PN_LIST = ['hc360', 'huangye88', 'pangwo', 'coovee', 'liebiao', 'china', 'ic114', 'cecb2b', 'trustexporter',
                    'x11467', 'x912688', 'zgtghccl', 'herostart', 'testmart', 'cc', 'eastsoo', 'x51sole', 'eb80', 'x258',
                    'x11467', 'x912688', 'zgtghccl', 'herostart', 'testmart', 'cc', 'eastsoo', 'x51sole', 'eb80', 'x258',
-                   'taojindi', 'd17', 'etlong', 'mfqyw', 'pe168', 'qymgc', 'atobo', 'czvv', 'qic']
+                   'taojindi', 'd17', 'etlong', 'mfqyw', 'pe168', 'qymgc', 'atobo', 'czvv', 'qic', 'mmic']
 
 
 
 
 def get_env():
 def get_env():