def merge_features(arr, pattern, func, do_remove):
feats = ut.regex_filter(arr.dtype.names, pattern)
print "features matched:", feats
if len(feats) > 1:
merged = [func(tuple(i)) for i in arr[feats]]
name = pattern + '_' + func.__name__
removefeats = set(feats) if do_remove else set([])
keepdtypes = [dt for dt in arr.dtype.descr if not dt[0] in removefeats]
newdtype = np.dtype(keepdtypes + [(name, arr[feats[0]].dtype)])
newarr = np.empty(arr.shape, dtype=newdtype)
for field in [d[0] for d in keepdtypes]:
newarr[field] = arr[field]
newarr[name] = merged
return newarr
else:
print "Not enough features to merge."
return arr
def regex_cols(arr, pattern):
return arr[['id1','id2','hit'] + ut.regex_filter(arr.dtype.names, pattern)]
