writelines() is faster than write() if your data can fit in RAM

filter_file.py

def filter_file(path, sieve):
     with open(path, 'r') as fr:
         lines = fr.readlines()
         with tempfile.NamedTemporaryFile(mode='w', delete=False) as temp:
             temp.writelines(l for l in lines if l.rstrip() not in sieve)
             temp.flush()
             os.fsync(temp.fileno())
     os.replace(temp.name, path)

@>>> timeit.timeit("filter_file('/tmp/t', ['abcnewsvideo 9758031'])", number=100, setup="from __main__ import filter_file")
20.872984636982437
@>>> timeit.timeit("filter_file('/tmp/t', ['abcnewsvideo 9758031'])", number=100, setup="from __main__ import filter_file")
20.709978723025415

def filter_file(path, sieve):
     with open(path, 'r') as fr:
         lines = fr.readlines()
         with tempfile.NamedTemporaryFile(mode='w', delete=False) as temp:
             for line in lines:
                 if line.rstrip() in sieve:
                     continue
                 temp.write(line)
             temp.flush()
             os.fsync(temp.fileno())
     os.replace(temp.name, path)
 
@>>> timeit.timeit("filter_file('/tmp/t', ['abcnewsvideo 9758031'])", number=100, setup="from __main__ import filter_file")
29.54833430200233
@>>> timeit.timeit("filter_file('/tmp/t', ['abcnewsvideo 9758031'])", number=100, setup="from __main__ import filter_file")
29.225350995984627