BioConductor source download script

From Biolanguage.org

BioConductor

 

#!/usr/bin/python
 
import urllib, re, os
 
BioCv = '2.3'
 
# Create the folder where the source will be downloaded
folder = 'SourceBioC-'+BioCv
if not os.path.exists(folder):
	os.mkdir(folder)
mainFolder = os.getcwd()
 
# Get the name of all the packages
source = 'https://readonly:readonly@hedgehog.fhcrc.org/bioconductor/trunk/madman/Rpacks/'
f = urllib.urlopen(source)
s = f.read()
f.close()
 
# Extract the name and put them in a list
tableName = []
s = s.split('\n')
for line in s:
	nameMotif = re.compile('href="(.*)"')
	try:
		name = nameMotif.findall(line)[0]
		name = name.replace('/',)
		if not name.startswith('http')  and name != '..':
			tableName.append(name)
	except:
		None
 
 
print '----------------'
print "There is %s library found" % len(tableName)
print '----------------'
 
i = 0
# For each go to the page of the release and download them
for lib in tableName:
	i = i + 1
	source = 'http://bioconductor.org/packages/%s/bioc/html/%s.html' % (BioCv, lib)
	#print source
	f = urllib.urlopen(source)
	s = f.read()
	f.close()
	s = s.split('\n')
 
	# Change folder and retrieve the tarball
	os.chdir(folder)
	for line in s:
		sourceMotif = re.compile('href=".*/(.*.tar.gz)"')
		try:
			source = sourceMotif.findall(line)[0]
			source = 'http://bioconductor.org/packages/%s/bioc/src/contrib/%s' % (BioCv, source)
			cmd = 'wget %s --quiet ' % source
			#print cmd
			try:
				print "Retrieve package %s on %s" % (i, len(tableName))
				print " " + lib
				os.system(cmd)
			except:
				print "Can not retrieve package: %s" % lib
		except:
			None
	os.chdir(mainFolder)
	print "----------------\n"