From 9fdbf60165db0400c2e8e6be2dc6e88138ac719a Mon Sep 17 00:00:00 2001
From: zmc <zmc_li@foxmail.com>
Date: 星期五, 22 十二月 2023 13:28:11 +0800
Subject: [PATCH] Merge branch 'master' of ssh://114.215.109.124:29418/fume-manage-python
---
src/Crawling.py | 41 ++++++++++-------------------------------
1 files changed, 10 insertions(+), 31 deletions(-)
diff --git a/src/Crawling.py b/src/Crawling.py
index 448d8db..fa332bd 100644
--- a/src/Crawling.py
+++ b/src/Crawling.py
@@ -120,7 +120,8 @@
count_all=count_all+get_OnePage(i,count)
if len(list_temp)==0: #濡傛灉璇ラ〉涓虹┖锛屽垯琛ㄧず璇ラ〉鍚庨潰閮芥棤鏁版嵁 閫�鍑哄惊鐜�
print('鍚庨潰椤垫暟涓虹┖锛岀埇鍘讳笅涓�涓簵閾�')
- break #閫�鍑哄惊鐜�
+ break
+ #閫�鍑哄惊鐜�
list_all.extend(list_temp) #灏嗕竴椤垫暟鎹垪琛ㄨ拷鍔犲埌list_all涓�
print("鐖彇浜嗙",page,"椤�")
page=page+1
@@ -604,11 +605,7 @@
global ck
global list_temp #浣跨敤鍏ㄥ眬鍙橀噺
list_temp.clear() #娓呯┖涓存椂琛�
- # session.headers = {
- # # 姝ゅ娉ㄦ剰cookie锛岃鑷繁鎶撳彇
- # # "Cookie":ck,
- # "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/111.0.0.0 Safari/537.36",
- # }
+
r = session.get(url, verify=False).text
soup = bs(r,'html.parser')
@@ -686,11 +683,7 @@
global list_temp #浣跨敤鍏ㄥ眬鍙橀噺
list_temp.clear() #娓呯┖涓存椂琛�
- # session.headers = {
- # # 姝ゅ娉ㄦ剰cookie锛岃鑷繁鎶撳彇
- # # "Cookie":ck,
- # "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/113.0.0.0 Safari/537.36",
- # }
+
r = session.get(url, verify=False).text
soup = bs(r,'html.parser')
@@ -707,7 +700,6 @@
print('鍒犻櫎鐗规畩鐨勫悗涓や釜')
print(list1)
list.append(list1)
- #list1.clear()
#print(list)
list_data=[]
@@ -753,36 +745,27 @@
urls=url_more() #杩斿洖鏂囦欢涓墍鏈夊簵閾虹殑url,甯︽渶澶ф樉绀烘潯鏁�100 銆倁rls鏄垪琛�
#print(urls)
teshu_url=[]
- #'shop=%25E9%25A3%259F%25E5%2585%25B6%25E5%25AE%25B6' 椋熷叾瀹�
special_url=['shop=%25E4%25BA%25BA%25E7%2594%259F%25E4%25B8%2580%25E4%25B8%25B2','shop=%25E7%25BC%2598%25E5%25AE%25B6','shop=%25E4%25B8%25B0%25E8%258C%2582%25E7%2583%25A4%25E4%25B8%25B2','shop=%25E6%25B3%25B0%25E7%2585%258C%25E9%25B8%25A1','shop=%25E5%25B0%258F%25E9%2593%2581%25E5%2590%259B']
for url in urls: #閬嶅巻鎵�鏈夊簵閾虹殑url
begin=url.find('&')+1
end=url.rfind('&')
- #print(begin,end)
#鎵惧埌鐗规畩鐨剈rl,杩涜鐗规畩澶勭悊
if url[begin:end] in special_url:
print('鍙戠幇鐗规畩鐨勶紒')
- already_spider_shopnum += 1 #鐖幓鐨勫簵閾烘暟閲忓姞1
+ already_spider_shopnum += 1 # 鐖幓鐨勫簵閾烘暟閲忓姞1
teshu_url.append(url)
#鍐嶅垹闄や竴涓垪琛ㄥ厓绱�
- url_teshu=url_add_time(url,date_begin,date_end) #缁欐墍鏈塽rl鍔犱笂鏃ユ湡
- list_to_MySql=get_MorePages_teshu(url_teshu,page) #搴楅摵鐨剈rl,鐖彇鐨勯〉鏁�
- # a=remove_Duplicates_list(list_to_MySql)
- # print('\n')
- # for item in a:
- # print(item)
+ url_teshu=url_add_time(url,date_begin,date_end) # 缁欐墍鏈塽rl鍔犱笂鏃ユ湡
+ list_to_MySql=get_MorePages_teshu(url_teshu,page) # 搴楅摵鐨剈rl,鐖彇鐨勯〉鏁�
+
if len(list_to_MySql) == 0 :
print('璇ュ搴楅摵鏃堕棿娈垫棤鏁版嵁锛屽凡璺宠繃')
continue
has_remove_duplicates = remove_Duplicates_list(list_to_MySql) #鍘婚櫎鏌愪釜搴楅摵鎸囧畾椤垫暟鎵�鏈夐噸澶嶇殑鏁版嵁
for item in has_remove_duplicates:
all_data.append(item)
- # is_minutes_exceed_30(has_remove_duplicates,con) # 灏嗘寚瀹氶〉鏁扮殑璁惧鏁呴殰鏁版嵁鍐欏叆鏁版嵁搴撳紓甯歌〃涓�
- # isExceeding(has_remove_duplicates,con) # 灏嗘寚瀹氶〉鏁版暟鎹啓鍏ユ暟鎹簱瓒呮爣琛ㄤ腑
- # ea_t_dev(has_remove_duplicates,con) # 灏嗘寚瀹氶〉鏁版暟鎹啓鍏ユ暟鎹簱璁惧淇℃伅琛ㄤ腑
- # fd_t_minbute(has_remove_duplicates,con) #灏嗘寚瀹氶〉鏁版暟鎹啓鍏ユ暟鎹簱鍒嗛挓鏁版嵁琛ㄤ腑
-
+
list_to_MySql.clear()
if url[begin:end]=='shop=%25E9%25A3%259F%25E5%2585%25B6%25E5%25AE%25B6': #椋熷叾瀹�
print('鍙戠幇鐗规畩鐨勶紒')
@@ -800,11 +783,7 @@
has_remove_duplicates = remove_Duplicates_list(list_to_MySql) #鍘婚櫎鏌愪釜搴楅摵鎸囧畾椤垫暟鎵�鏈夐噸澶嶇殑鏁版嵁
for item in has_remove_duplicates:
all_data.append(item)
- # is_minutes_exceed_30(has_remove_duplicates,con) # 灏嗘寚瀹氶〉鏁扮殑璁惧鏁呴殰鏁版嵁鍐欏叆鏁版嵁搴撳紓甯歌〃涓�
- # isExceeding(has_remove_duplicates,con) # 灏嗘寚瀹氶〉鏁版暟鎹啓鍏ユ暟鎹簱瓒呮爣琛ㄤ腑
- # ea_t_dev(has_remove_duplicates,con) # 灏嗘寚瀹氶〉鏁版暟鎹啓鍏ユ暟鎹簱璁惧淇℃伅琛ㄤ腑
- # fd_t_minbute(has_remove_duplicates,con) #灏嗘寚瀹氶〉鏁版暟鎹啓鍏ユ暟鎹簱鍒嗛挓鏁版嵁琛ㄤ腑
-
+
list_to_MySql.clear()
for t in teshu_url: #浠巙rls琛ㄤ腑鍒犻櫎鐗规畩鐨�
urls.remove(t)
--
Gitblit v1.9.3