Difference between
modified post 24c by Anonymous on Thu 6th Jan 2011 18:48
original post fea41 by Anonymous on Thu 6th Jan 2011 16:46
Download diff
Show old version | new version | both versions

import urllib2 as url

1		-	import urllib2 as url
	1	+	import urllib2
2	2
3	3		def getBetween(s, e, searched):
4	4		sInd = searched.find(s)
5	5		return searched[sInd + len(s):searched.find(e, sInd)]
6	6
7	7		def get(url):
8	8		isHards = []
9	9		isbns = []
10	10		done = False
11	11		for pg in xrange(1, 5000):
12		-	html = url.urlopen(url + '?PG=%d' % pg).read()
	12	+	html = urllib2.urlopen(url + '?PG=%d' % pg).read()
13	13		eInd = 0
14	14		while True:
15	15		sInd = html.find("<span class='regGray11px'>(", eInd)
16	16		if sInd == -1:
17	17		if eInd == 0:
18	18		done = True
19	19		break
20	20		eInd = html.find("</span>\n", sInd)
21	21		foo = html[sInd + len("<span class='regGray11px'>("):eInd]
22	22		isHards.append('Hardback' in foo)
23	23		isbns.append(getBetween('ISBN10: ', ';', foo))
24	24		if done:
25	25		break
26		-	return results
	26	+	return isHards, isbns
27	27
28	28		print get('http://www.textbooks.com/Catalog/KMG/Geometry-for-K-12.php')
29	29
30	30		######
31	31
32	32		http://www.chartwellyorke.com/fathom/fathomdemos.html
33	33
34	34		http://isbndb.com/