get_mtID.py

import re
import time
from sys import argv


"""
snipet to extract mtIDs of a selected mutant line in medicago-mutant.noble.org
"""


from selenium import webdriver
from selenium.webdriver.common.keys import Keys

dic=dict()

p = re.compile(">(CHR\S+?)</a>")

mutants=open(argv[1])

for link in mutants:
    try:
        dic[link] = []
        url='http://medicago-mutant.noble.org/mutant/FNBline1.php?id='+link
        browser = webdriver.Firefox()
        browser.get(url)
        page=browser.page_source
        for i in p.findall(page):
            #print(i)
            dic[link].append(i)
        #time.sleep(5) # delays for 5 seconds to keep our good relationship with the server
        browser.quit()
    except:
		print('PROBLEM!')

for i in sorted(dic.keys()):
    print(i.strip())
    for a in dic[i]:
        print(a)