Merge pull request #25 from wuyue92tree/dev

Dev

Author: wuyue92tree

README.md

@@ -1,7 +1,7 @@
 # Crwy
 
 [![PyPI Version](https://img.shields.io/pypi/v/Crwy.svg)](https://pypi.python.org/pypi/Crwy)
-[![Build Status](https://travis-ci.org/wuyue92tree/crwy.svg?branch=1.1.2)](https://travis-ci.org/wuyue92tree/crwy)
+[![Build Status](https://travis-ci.org/wuyue92tree/crwy.svg?branch=1.1.3)](https://travis-ci.org/wuyue92tree/crwy)
 
 # 简介
 
@@ -34,7 +34,7 @@ pip install crwy
 ```
 
 or
-前往下载: https://pypi.python.org/pypi/Crwy/1.1.2/
+前往下载: https://pypi.python.org/pypi/Crwy/1.1.3/
 
 # 使用手册
 

crwy/VERSION

@@ -1 +1 @@
-1.1.2
+1.1.3

crwy/utils/extend/xunma.py

@@ -11,7 +11,7 @@
 """
 
 
-from __future__ import print_function
+from __future__ import print_function, unicode_literals
 
 from crwy.spider import Spider
 from crwy.exceptions import CrwyExtendException
@@ -38,7 +38,7 @@ def login(self):
                                                        password=self.password)
             res = self.html_downloader.download(url)
 
-            return res.content.strip().split("&")[0]
+            return res.text.strip().split("&")[0]
         except Exception as e:
             raise CrwyExtendException(e)
 
@@ -58,11 +58,7 @@ def get_phone(self, token, phone_type='', phone=''):
                                          phone_type=phone_type, phone=phone)
 
             res = self.html_downloader.download(url)
-            # if 'success' not in res.content:
-            #     raise MfExtendException("[XunMa] get phone failed.")
-            #
-            # # print(res.content)
-            return res.content.strip().split(';')[0]
+            return res.text.strip().split(';')[0]
 
         except Exception as e:
             raise CrwyExtendException(e)
@@ -82,7 +78,7 @@ def get_message(self, token, phone):
                                       item_id=self.item_id, phone=phone)
             res = self.html_downloader.download(url)
 
-            return res.content.strip().split('&')[-1]
+            return res.text.strip().split('&')[-1]
 
         except Exception as e:
             raise CrwyExtendException(e)

crwy/utils/extend/yima.py

@@ -10,7 +10,7 @@
 这一行开始写关于本文件的说明与解释
 """
 
-from __future__ import print_function
+from __future__ import print_function, unicode_literals
 
 from crwy.spider import Spider
 from crwy.exceptions import CrwyException
@@ -39,10 +39,10 @@ def login(self):
                                                 password=self.password)
             res = self.html_downloader.download(url)
 
-            if 'success' not in res.content:
+            if 'success' not in res.text:
                 raise CrwyException("[YiMa] Login failed.")
 
-            return res.content.strip().split("|")[-1]
+            return res.text.strip().split("|")[-1]
         except Exception as e:
             raise CrwyException(e)
 
@@ -65,11 +65,11 @@ def get_phone(self, token, phone_type='',
                                           phone_type=phone_type, phone=phone)
 
             res = self.html_downloader.download(url)
-            if 'success' not in res.content:
+            if 'success' not in res.text:
                 raise CrwyException("[YiMa] get phone failed.")
 
-            # print(res.content)
-            return res.content.strip().split('|')[-1]
+            # print(res.text)
+            return res.text.strip().split('|')[-1]
 
         except Exception as e:
             raise CrwyException(e)
@@ -88,11 +88,11 @@ def get_message(self, token, phone):
                                       phone=phone)
             res = self.html_downloader.download(url)
 
-            if 'success' not in res.content:
+            if 'success' not in res.text:
                 raise CrwyException("[YiMa] get message failed.")
 
             else:
-                return res.content.strip().split('|')[-1]
+                return res.text.strip().split('|')[-1]
 
         except Exception as e:
             raise CrwyException(e)
@@ -105,7 +105,7 @@ def release_phone(self, token, phone):
                                       phone=phone)
             res = self.html_downloader.download(url)
 
-            if 'success' not in res.content:
+            if 'success' not in res.text:
                 raise CrwyException("[YiMa] release phone failed.")
 
         except Exception as e:
@@ -119,7 +119,7 @@ def add_black(self, token, phone):
                                       phone=phone)
             res = self.html_downloader.download(url)
 
-            if 'success' not in res.content:
+            if 'success' not in res.text:
                 raise CrwyException("[YiMa] black phone failed.")
 
         except Exception as e:

crwy/utils/html/font_analysis.py

@@ -86,7 +86,7 @@ def analysis(self, is_clean=True, debug=False):
 
         self.save_font()
         self.get_font_xml()
-        with open(self.xml_path, 'r') as xml:
+        with open(self.xml_path, 'rb') as xml:
             soups = self.html_parser.parser(xml.read())
             ttglyph_lst = soups.find('glyf').find_all('ttglyph')[1:]
             analysis_res = []

crwy/utils/no_sql/redis_m.py

@@ -24,9 +24,8 @@
 class RedisDb(object):
     def __init__(self, **kwargs):
         if 'url' in kwargs.keys():
-            url = kwargs.get('url')
-            db = kwargs.get('db', 0)
-            self.pool = redis.ConnectionPool.from_url(url, db=db, **kwargs)
+            url = kwargs.pop('url')
+            self.pool = redis.ConnectionPool.from_url(url, **kwargs)
         else:
             self.pool = redis.ConnectionPool(**kwargs)
         self.db = redis.StrictRedis(connection_pool=self.pool)

crwy/utils/scrapy_plugs/dupefilters.py

@@ -69,6 +69,10 @@ def request_seen(self, request):
         if not request.meta.get('dupefilter_key', None):
             return False
 
+        if len(request.meta.get('redirect_urls', [])) > 0:
+            # skip url from redirect
+            return False
+
         dupefilter_key = request.meta.get('dupefilter_key')
         dupefilter_key = hashlib.sha1(dupefilter_key).hexdigest() if \
             self.do_hash else dupefilter_key

crwy/utils/scrapy_plugs/middlewares.py

@@ -65,7 +65,10 @@ def process_request(self, request, spider):
                     'no user in cookie_pool:{}'.format(self.site))
         cookie = self.h.hget(user)
         if cookie:
-            request.cookies = json.loads(cookie)
-            spider.logger.debug('get_cookie_success: {}'.format(user))
+            # 字典存入redis，取出时未string，通过eval进行还原
+            request.cookies = eval(cookie)
+            spider.logger.debug('get_cookie_success: {}'.format(
+                user.decode('utf-8')))
         else:
-            spider.logger.warning('get_cookie_failed: {}'.format(user))
+            spider.logger.warning('get_cookie_failed: {}'.format(
+                user.decode('utf-8')))