histogram.py errors if all values are equal #32

Activity

1. havanagrawal commented on Apr 20, 2016

   

   havanagrawal

   on Apr 20, 2016

   More actions

   Hi, I'd like to fix this issue, please let me know how I can contribute
   :)
2. shabbybanks commented on Nov 7, 2016

   

   shabbybanks

   on Nov 7, 2016

   More actions

   This error is also hit in the corner case where there is only one value input.
3. shabbybanks commented on Nov 7, 2016

   

   shabbybanks

   on Nov 7, 2016

   More actions

   The following patch works for me, but is a little wonky in terms of Decimal math:

   --- histogram.py  2016-11-04 14:54:19.092902000 -0400
   +++ fixed_histogram.py  2016-11-07 17:35:01.707723906 -0500
   @@ -139,9 +139,16 @@
            max_v = max(data, key=lambda x: x.value)
            max_v = max_v.value
   
   -    if not max_v > min_v:
   -        raise ValueError('max must be > min. max:%s min:%s' % (max_v, min_v))
   -    diff = max_v - min_v
   +    fake_max = max_v
   +    fake_min = min_v
   +
   +    if not fake_max > fake_min:
   +        print >>sys.stderr, 'max must be > min. max:%s min:%s' % (max_v, min_v)
   +        fake_min = max(fake_min - Decimal("0.5"),Decimal("0.5") * fake_min)
   +        fake_max = min(fake_max + Decimal("0.5"),Decimal("2.0") * fake_max)
   +
   +
   +    diff = fake_max - fake_min
   
        boundaries = []
        bucket_counts = []
   @@ -152,18 +159,18 @@
            bound_sort = sorted(map(Decimal, bound))
   
            # if the last value is smaller than the maximum, replace it
   -        if bound_sort[-1] < max_v:
   -            bound_sort[-1] = max_v
   +        if bound_sort[-1] < fake_max:
   +            bound_sort[-1] = fake_max
   
            # iterate through the sorted list and append to boundaries
            for x in bound_sort:
   -            if x >= min_v and x <= max_v:
   +            if x >= fake_min and x <= fake_max:
                    boundaries.append(x)
   -            elif x >= max_v:
   -                boundaries.append(max_v)
   +            elif x >= fake_max:
   +                boundaries.append(fake_max)
                    break
   
   -        # beware: the min_v is not included in the boundaries, so no need to do a -1!
   +        # beware: the fake_min is not included in the boundaries, so no need to do a -1!
            bucket_counts = [0 for x in range(len(boundaries))]
            buckets = len(boundaries)
        else:
   @@ -173,7 +180,7 @@
            step = diff / buckets
            bucket_counts = [0 for x in range(buckets)]
            for x in range(buckets):
   -            boundaries.append(min_v + (step * (x + 1)))
   +            boundaries.append(fake_min + (step * (x + 1)))
   
        skipped = 0
        samples = 0
   @@ -185,7 +192,7 @@
                mvsd.add(record.value, record.count)
                accepted_data.append(record)
            # find the bucket this goes in
   -        if record.value < min_v or record.value > max_v:
   +        if record.value < fake_min or record.value > fake_max:
                skipped += record.count
                continue
            for bucket_postion, boundary in enumerate(boundaries):
   @@ -203,8 +210,8 @@
        if options.mvsd:
            print "# Mean = %f; Variance = %f; SD = %f; Median %f" % (mvsd.mean(), mvsd.var(), mvsd.sd(), median(accepted_data, key=lambda x: x.value)
   )
        print "# each ∎ represents a count of %d" % bucket_scale
   -    bucket_min = min_v
   -    bucket_max = min_v
   +    bucket_min = fake_min
   +    bucket_max = fake_min
        percentage = ""
        format_string = options.format + ' - ' + options.format + ' [%6d]: %s%s'
        for bucket in range(buckets):