您的位置:首页 > 编程语言 > Python开发

python之扫描本地信息到Solr服务

2015-11-19 00:18 489 查看
##scan_to_solr.py

#-*-coding:utf-8-*-
import urllib
import os,sys,re
import urllib2,httplib
import threading
import time
urlMap = {}
gEnd = False
gExit = False
gOK = 0
gErr = 0
reload(sys)
#sys.setdefaultencoding('utf-8')
sys.setdefaultencoding('gbk')
gCount = 0
gErrorList = []
gErrNumber = 0
def getUrl(url):
try:
return urllib2.urlopen(url,timeout=240).read()
except:
print "XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXload error~"
return None

def addIndex(key,value):
url = u'http://192.168.4.60/baidu/src/addServer.php?ch=%s&jp=%s' % (key.decode("gbk"),value)
ret = getUrl(url.encode("utf-8"))
if ret != None:
return True
return False

def getList(dirname,pFunc):
try:
ls=os.listdir(dirname)
except:
print dirname,'is access deny'
else:
for file in ls:
temp = os.path.join(dirname,file)
if(os.path.isdir(temp)):
getList(temp,pFunc)
else:
pFunc(dirname,file)

"""
callBack Function
"""
def doAddFile(dir,file):
global gCount
global urlMap,gOK,gErr,gErrNumber,gErrorList
if file == "entries" or file == "toIndex.py" or file == "index.txt":
return
if re.search("(.ppt|.pptx|.xls|.txt|.doc|.vsd|.pdf|.rar|.xlsx|.zip|.docx|.tar|.gz|.iso|.rpm|.sql)$",file):
gCount = gCount + 1
dir = dir.replace("\\","/")
svnUrl = 'svn://192.168.0.248:9997/DOC/' + u'软件部/'
anchor = "%s%s" % (svnUrl,dir[11:].decode("gbk"))
print file
file = file.replace(" ","")
print "###############"
print file
print "###############(%d)" % (gCount)
ret = addIndex(file,anchor)
if ret == True:
gOK = gOK + 1
print "OOO KKK(%d)\n" % (gOK)
time.sleep(1)
else:
gErr = gErr + 1
print "NNN GGG(%d)\n" % (gErr)
addIndex(key,value)

"""
Main Inport
"""
print "\n============================== Begin ScanDir ========================================"
getList(os.getcwd(),doAddFile)
print "ScanFile(%d)" % (gCount)
print "\n========================================================================================"
print "OK(%d) ,Error(%d)" % (gOK, gErr)


@@用法说明

scan_to_solr.py

扫描本地文件提交到Solr的模糊检索服务中
内容来自用户分享和网络整理,不保证内容的准确性,如有侵权内容,可联系管理员处理 点击这里给我发消息
标签: