Ubuntu Wiki

   1 #Python Version 2.5
   2 #
   3 #
   4 #we need the following modules
   5 
   6 import urllib2,string
   7 from BeautifulSoup import BeautifulSoup
   8 
   9 
  10 def main():
  11     # parameter and constants
  12     team_url='https://launchpad.net/~ubuntu-co/+members'
  13 
  14     # open team's web page
  15     team_page=urllib2.urlopen(team_url)
  16 
  17     # Use BeautifulSoup to parse html
  18     soup0 = BeautifulSoup(team_page)
  19     
  20     #Find active members
  21     p=soup0.find("table",{ "id":"activemembers"})
  22     
  23     #Convert element to string and use BeautifulSoup again
  24     soup1 = BeautifulSoup(str(p))
  25     
  26     #Prepare output as a table
  27     members_table=['<table class="listing sortable" id="activemembers"><thead><tr><th>Nombre</th><th>Ubuntero?</th><th>Karma</th></tr></thead><tbody>']
  28     
  29     #Find all the ocurrences of <a href=""> in the member's table
  30     member_data=soup1.findAll('a')
  31     
  32     for i in member_data:
  33 	# get member's url
  34 	member_url='https://launchpad.net'+str(i.get('href'))
  35         # Open member's web page
  36 	member_page=urllib2.urlopen(member_url)
  37 	# Use BeautifulSoup to parse html
  38 	soup2=BeautifulSoup(member_page)
  39         #Check to see if the opened page belongs to a group, if it's not the case, get relevant data
  40         q=soup2.find("a",{ "class":"menu-link-members"})
  41 	if q == None:
  42 	        # add member data to output table
  43         	members_table.append('<tr><td>'+str(i).replace('href="','href="https://launchpad.net')+'</td><td>')
  44 	        #find the word "Ubuntero" and if it says "Yes" or no, add relevant data to output table
  45 		u=soup2.find(text='Ubuntero:')
  46 		if u <> None:
  47 			v=u.findNext(text=True)
  48 			if v.find("Yes") > 0:
  49 	   			members_table.append('Si')
  50 			else:	 
  51 		   		members_table.append('No')
  52 		else:	
  53 			members_table.append('No')
  54 		# Find Member's karma, add it to output table
  55 		karma=soup2.find("span",{ "id":"karma-total"}).contents[0]
  56 		members_table.append('</td><td>'+str(karma)+'</td></tr>')
  57     # Add trailing info for table	
  58     members_table.append('</tbody></table>')
  59     # Convert element to string
  60     doc= ''.join(members_table)
  61     # Use BeautifulSoup to parse html
  62     soup = BeautifulSoup(doc)
  63     # Print output table
  64     print soup.prettify()
  65       
  66 #call main function
  67 main()