Welcome, guest | Sign In | My Account | Store | Cart

I wrote this method to be, first and foremost, flexible. This method implements what is referred to as a Schwartzian transform algorithm. This method does provide a mechanism to ensure stability in the sort. This version does a sort in place.

Python, 63 lines
 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63``` ```def sortByAttrs(seq, attrs): listComp = ['seq[:] = [('] for attr in attrs: listComp.append('seq[i].%s, ' % attr) listComp.append('i, seq[i]) for i in xrange(len(seq))]') exec('%s' % ''.join(listComp)) seq.sort() seq[:] = [obj[-1] for obj in seq] return # # begin test code # from random import randint class a: def __init__(self): self.x = (randint(1, 5), randint(1, 5)) class b: def __init__(self): self.x = randint(1, 5) self.y = (a(), a()) class c: def __init__(self, arg): self.x = arg self.y = b() if __name__ == '__main__': aList = [c(1), c(2), c(3), c(4), c(5), c(6)] print '\n...to be sorted by obj.y.y[1].x[1]' print ' then, as needed, by obj.y.x' print ' then, as needed, by obj.x\n\n ', for i in range(6): print '(' + str(aList[i].y.y[1].x[1]) + ',', print str(aList[i].y.x) + ',', print str(aList[i].x) + ') ', sortByAttrs(aList, ['y.y[1].x[1]', 'y.x', 'x']) print '\n\n...now sorted by listed attributes.\n\n ', for i in range(6): print '(' + str(aList[i].y.y[1].x[1]) + ',', print str(aList[i].y.x) + ',', print str(aList[i].x) + ') ', print # # end test code # ```

As demonstrated in the test code above, a list can be sorted by attribute, by dotted attribute, and by dotted attribute[index] forms.

The list/tuple of passed attributes can have an arbitrary length greater than zero. This allows for a very robust determination of comparison 'on the fly'. If an object in the list does not have a passed attribute, the method will throw (and not catch) an AttributeError.

 Created by Frank H on Thu, 18 Sep 2003 (PSF)