17 fr1 = open(file1).readlines()
18 fr2 = open(file2).readlines() ###如果是fr2 = open(file2) ,那么for 嵌套循环会报错
19 sh_dict = {}
20 cnt_mil = 0
21 cnt_ent =0
22 cnt_tech = 0
23 ch1 = ''
24 for line in fr1:
25 #print 'll'
26 url = line.split('\t')[0]
27 p1 = urlparse.urlparse(url)[2]
28 for line2 in fr2:
29 #print 'ii'
30 #print '#'+p1+'#'+urlparse.urlparse(line2.split('\t')[0])[2]+'#'
31 #print urlparse.urlparse(line2.split('\t')[0])[2]
32 if p1 == urlparse.urlparse(line2.split('\t')[0])[2]:
33 ch1 = getchannel(url)
34 #print ch1
35 if ch1 in ['jmqmil','mil'] and cnt_mil <34:
36 print line2.strip()
37 cnt_mil +=1
18 fr2 = open(file2).readlines() ###如果是fr2 = open(file2) ,那么for 嵌套循环会报错
19 sh_dict = {}
20 cnt_mil = 0
21 cnt_ent =0
22 cnt_tech = 0
23 ch1 = ''
24 for line in fr1:
25 #print 'll'
26 url = line.split('\t')[0]
27 p1 = urlparse.urlparse(url)[2]
28 for line2 in fr2:
29 #print 'ii'
30 #print '#'+p1+'#'+urlparse.urlparse(line2.split('\t')[0])[2]+'#'
31 #print urlparse.urlparse(line2.split('\t')[0])[2]
32 if p1 == urlparse.urlparse(line2.split('\t')[0])[2]:
33 ch1 = getchannel(url)
34 #print ch1
35 if ch1 in ['jmqmil','mil'] and cnt_mil <34:
36 print line2.strip()
37 cnt_mil +=1