import sys
contols = sys.argv[1]
downloads = sys.argv[2:]

contols = open(contols).readlines()
# 457dda4157df483db29be7378e934fff  Clean/1/V300036655_L3_AE00917519-1_1.fq.gz

downloads = [list(set(open(d).readlines())) for d  in downloads]
downloads = sum(downloads, [])
# 71c8e5831cc8aee698d8ffb89b5b329d 4/1_V300036655_L2_AE17428503-2_1.fq.gz

total = 0
for i in contols:
    i = i.strip()
    if len(i)==0:
        continue
    md5, fname = i.split()
    fname = fname.replace("Clean/","")
    if not fname[0] in ["1","4","5","7"]:
        continue
    found = [q for q in downloads if q.startswith(md5)]
    if len(found)>1:
        print ("Multiple md5 files: ",found)
    if len(found) != 1:
        print ("File ",i," not found")
    total += 1
print ("Processed ",total," files")
