add clear ranks_sum and dup for nan cases

Author: peterpanmj

pandas/_libs/algos_rank_helper.pxi.in

@@ -44,6 +44,10 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
 
  {{if dtype == 'object'}}
  ndarray sorted_data, values
+ ndarray[np.uint8_t, cast=True] sorted_namask
+ {{elif dtype != 'uint64'}}
+ ndarray[{{ctype}}] sorted_data, values
+ ndarray[np.uint8_t, cast=True] sorted_namask
  {{else}}
  ndarray[{{ctype}}] sorted_data, values
  {{endif}}
@@ -54,7 +58,7 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
  {{if dtype == 'uint64'}}
  {{ctype}} val
  {{else}}
- {{ctype}} val, nan_value
+ {{ctype}} val, nan_value, isnan
  {{endif}}
 
  float64_t sum_ranks = 0
@@ -123,17 +127,23 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
  _as = _as[::-1]
 
  sorted_data = values.take(_as)
+ # need to distinguish between pos/neg nan and real nan when keep_na is true
+ {{if dtype != 'uint64'}}
+ sorted_namask = mask.take(_as)
+ sorted_namask = sorted_namask.astype(np.bool)
+ {{endif}}
  argsorted = _as.astype('i8')
 
  {{if dtype == 'object'}}
  for i in range(n):
  sum_ranks += i + 1
  dups += 1
-
+ isnan = sorted_namask[i]
  val = util.get_value_at(sorted_data, i)
 
- if (val is nan_value) and keep_na:
+ if isnan and keep_na:
  ranks[argsorted[i]] = nan
+ sum_ranks = dups = 0
  continue
 
  count += 1.0
@@ -168,8 +178,10 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
  val = sorted_data[i]
 
  {{if dtype != 'uint64'}}
- if (val == nan_value) and keep_na:
+ isnan = sorted_namask[i]
+ if isnan and keep_na:
  ranks[argsorted[i]] = nan
+ sum_ranks = dups = 0
  continue
  {{endif}}