練習

gethtml-.pyで保存

# gethtml- http://...  => normal
# gethtml- - => run with file (='list.txt', url list)
# save.name = 'http_--sample.net-_200709120808.htm' etc.

import sys
import urllib2
import time
from datetime import date
import string

def strr(d,le):
	return ('0'+str(d))[-le:]

def timestr(ti):
	return strr(ti[0],4)+strr(ti[1],2)+strr(ti[2],2)+strr(ti[3],2)+strr(ti[4],2)
	# +strr(ti[5],2)

def nowt():
	return timestr(time.localtime())

def escape(f):
	return f.replace('/','-').replace(':','_').replace('?','+')

def getd(url):
	sf=escape(url)
	tof=sf+'_'+nowt()+".htm"
	print tof
	o=open(tof,'w+')
	
	for line in urllib2.urlopen(url):
	  #print line,
	  o.write(line)
	o.close

url=sys.argv[1]
if url=='-' :
	inf=open('list.txt')
	li=[]
	for x in inf.readlines():
		li.append(string.strip(x))
	for url in li:
		getd(url)
	sys.exit()

getd(url)