def updateState(self, url, state):
db = self._connect()
if (not isinstance(url, dict)):
id = md5(url)
else:
id = md5(url["url"])
db[self._tbName].update({"_id": id}, {"$set": {"state": state}})
def updateState(self, url, state):
conn = sqlite3.connect(self._dbPath)
try:
if(not isinstance(url,dict)):
id = md5(url)
else:
id = md5(url["url"])
conn.cursor().execute("update urls set state={} where id='{}'".format(state, id))
conn.commit()
except Exception as err:
self.log(err)
conn.close()
def resetUrls(self, urls):
db = self._connect()
for url in urls:
if (not isinstance(url, dict)):
id = md5(url)
url = {'url': url, '_id': id, 'state': 0}
else:
id = md5(url["url"])
url['_id'] = id
url['state'] = 0
if (not self.checkUrl(url["url"], None)):
db[self._tbName].insert(url)
else:
db[self._tbName].update({"_id": url["_id"]},
{"$set": {
"state": 0
}})
def resetUrls(self, urls):
datas=[]
for url in urls:
if(not isinstance(url,dict)):
id = md5(url)
url={'url':url}
else:
id = md5(url["url"])
datas.append((id,json.dumps(url),getTimeNow()))
if not len(datas): return
conn = sqlite3.connect(self._dbPath)
try:
cursor = conn.cursor()
# tmp = tuple(datas)
cursor.executemany("REPLACE into urls(id,json,state,tm) values(?,?,0,?)",tuple(datas))
conn.commit()
except Exception as err:
self.log(err)
conn.close()
def checkUrl(self, url, i):
if not self._duplicateRemoval: return False
db = self._connect()
id = md5(url)
tbName = self._tbName
if (i): tbName = tbName + str(i)
url = db[tbName].find_one({"_id": id})
if (not url and not i):
url = db[self._tbName].find_one({"_id": id})
return url
def checkUrl(self,url):
if not self._duplicateRemoval: return False
conn = sqlite3.connect(self._dbPath)
try:
flag=False
id = md5(url)
cursor = conn.cursor().execute("select id from urls where id='{}' limit 0,1".format(id))
for row in cursor:
flag = True
except Exception as err:
self.log(err)
conn.close()
return flag
def saveUrl(self, urls, i=None):
db = self._connect()
flag = False
for url in urls:
if (not isinstance(url, dict)):
id = md5(url)
url = {'url': url, '_id': id, 'state': 0}
else:
id = md5(url["url"])
url['_id'] = id
url['state'] = 0
if (i != 0):
i = self._getIndex(url["url"])
if (not self.checkUrl(url["url"], i)):
tbName = self._tbName
if (i):
tbName = tbName + str(i)
db[tbName].insert(url)
if (i in self._totalCount): self._totalCount[i] += 1
else: self._totalCount[i] = 1
flag = True
if (flag):
self._tbCache[tbName] = False
def _saveHtml(self, url, html):
suffix = '.htm'
index = url.rfind('/')
if (index > 10):
index = url.rfind('.', index)
if (index > 0 and len(url) - index < 10):
suffix = url[index:]
filename = md5(url) + suffix
if (not os.path.exists(self._htmlPath)):
os.mkdir(self._htmlPath)
file = io.open(self._htmlPath + filename, "w", encoding="utf-8")
file.write(html)
file.close()
return filename
def resetUrls(self, urls):
self._lock.acquire()
try:
flag=False
for url in urls:
if(not isinstance(url,dict)):
url={'url':url}
id=md5(url['url'])
self._urls.append(url)
self._dic.add(id)
self._writeFile(url,id)
flag=True
if(flag):
self._flushFile()
except Exception as err:
printInfo(err)
finally:
self._lock.release()
def saveUrl(self, urls,i=None):
# if (type(urls).__name__=='dict'):
# urls=urls["Urls"]
self._lock.acquire()
try:
flag=False
for url in urls:
if(not isinstance(url,dict)):
url={'url':url}
if(md5(url['url']) not in self._dic):
self._urls.append(url)
self._dic.add(md5)
self._writeFile(url,md5)
flag=True
if(flag):
self._flushFile()
except Exception as err:
printInfo(err)
finally:
self._lock.release()
def _saveHtml(self, url, html):
filename = md5(url) + '.htm'
file = io.open(self._htmlPath + filename, "w", encoding="utf-8")
file.write(html)
file.close()
def checkUrl(self,url): if not self._duplicateRemoval: return False _r = redis.Redis(connection_pool=self._pool) result = _r.sadd(self._setName, md5(url)) return not result
def checkUrl(self,url): return md5(url) in self._dic