Source

psilib / test_spider.py

Lynn Rees 2c86a90 









Lynn Rees 700da08 

Lynn Rees fd2482f 
Lynn Rees 2c86a90 
Lynn Rees 700da08 
Lynn Rees 2c86a90 
Lynn Rees 700da08 
Lynn Rees 2c1c6d3 
Lynn Rees fd2482f 
Lynn Rees 2c86a90 
Lynn Rees 700da08 
Lynn Rees 2c1c6d3 

Lynn Rees 5ee9b11 
Lynn Rees 9ba3572 
Lynn Rees 5ee9b11 
Lynn Rees 9ba3572 
Lynn Rees 2c86a90 






Lynn Rees fd2482f 
Lynn Rees 2c86a90 






Lynn Rees fd2482f 
Lynn Rees 2c1c6d3 


Lynn Rees 5ee9b11 
Lynn Rees 9ba3572 
Lynn Rees 5ee9b11 
Lynn Rees 9ba3572 
import spider
from pprint import pprint

if __name__ == '__main__':
    a = spider.ftpurls('ftp://localhost/')
    print 1; pprint(a)
    a = spider.ftppaths('ftp://localhost')
    print 2; pprint(a)
    a = spider.weburls('http://localhost/')
    print 3; pprint(a)
    a = spider.weburls('http://localhost/', 200, 5, 3)
    print 4; pprint(a)
    spider.ftpmirror('e:\\ftp\\', 14, 'ftp://localhost/')
    a = spider.ftpspider('ftp://localhost/')
    print 5; pprint(a)
    a = spider.webpaths('http://localhost/')
    print 6; pprint(a)
    spider.webreport('e:\\web1.txt', 'http://localhost/')
    spider.webmirror('e:\\web\\', 18, 'http://localhost/')
    a = spider.webspider('http://localhost/')
    print 7; pprint(a)
    spider.urlreport('e:\\web2.txt', 'http://localhost/',)
    spider.badurlreport('e:\\web3.txt', 'http://localhost/')
    spider.badhtmreport('e:\\web4.txt', 'http://localhost/')
    spider.redireport('e:\\web5.txt', 'http://localhost/')
    spider.outreport('e:\\web6.txt', 'http://localhost')
    spider.othereport('e:\\web7.txt', 'http://localhost/')
    a = spider.Spider('ftp://localhost/', 200, 16)
    a.ftppaths()
    print 1; pprint(a.paths)
    a.ftpurls()
    print 2; pprint(a.urls)
    a.ftpspider()
    print 3; pprint(a.urls); pprint(a.paths)
    a.ftpmirror('e:\\ftp\\', 7)
    a = spider.Spider('http://localhost/', 200, 16)
    a.weburls()
    print 4; pprint(a.urls)
    a.webpaths()
    print 5; pprint(a.paths)
    a.webspider()
    print 6; pprint(a.urls); pprint(a.paths)
    a.webmirror('e:\\web\\', 18)
    a.webreport('e:\\web1.txt')
    a.urlreport('e:\\web2.txt')
    a.badurlreport('e:\\web3.txt')
    a.badhtmreport('e:\\web4.txt')
    a.redireport('e:\\web5.txt')
    a.outreport('e:\\web6.txt')
    a.othereport('e:\\web7.txt')