コード例 #1
0
ファイル: crawl.py プロジェクト: dubstack/Predictube


ids = getId()

id_file = open('sqlite_id.txt','w')
for id in ids:
	id_file.write(id[0]+'\n')
id_file.close()





c = Crawler()
c._crawl_delay_time = 1.2
c._cookie_update_delay_time = 1
c.batch_crawl('sqlite_id.txt','/home/aayush/Desktop/assign/socialComputing/sqlite_data')

# ofile = open("youtube_data.csv",'w')
# errorFile = open("errorIds.csv",'w')
# writer = csv.writer(ofile)
# error_writer = csv.writer(errorFile)
# writer.writerow(['id','numShare','numSubscriber','watchTime','uploadDate','dailyViewCount'])

# c = Crawler()
# c._crawl_delay_time = 1 
# c._cookie_update_delay_time = 1

# for id in ids:
# 	i=0
コード例 #2
0
    con = sqlite3.connect('youtube-data-sqlite/youtube.sqlite')
    cur = con.cursor()
    sql = "SELECT id FROM youtube_statistics"
    ids = cur.execute(sql)
    return ids


ids = getId()

id_file = open('sqlite_id.txt', 'w')
for id in ids:
    id_file.write(id[0] + '\n')
id_file.close()

c = Crawler()
c._crawl_delay_time = 1.2
c._cookie_update_delay_time = 1
c.batch_crawl('sqlite_id.txt',
              '/home/aayush/Desktop/assign/socialComputing/sqlite_data')

# ofile = open("youtube_data.csv",'w')
# errorFile = open("errorIds.csv",'w')
# writer = csv.writer(ofile)
# error_writer = csv.writer(errorFile)
# writer.writerow(['id','numShare','numSubscriber','watchTime','uploadDate','dailyViewCount'])

# c = Crawler()
# c._crawl_delay_time = 1
# c._cookie_update_delay_time = 1

# for id in ids: