def readAndGroupTable( infile, options ):
"""read table from infile and group.
"""
fields, table = CSV.ReadTable( infile, with_header = options.has_headers, as_rows = True )
options.columns = getColumns( fields, options.columns )
assert options.group_column not in options.columns
converter = float
new_fields = [ fields[options.group_column] ] + [ fields[x] for x in options.columns ]
if options.group_function == "min":
f = min
elif options.group_function == "max":
f = max
elif options.group_function == "sum":
f = lambda z: reduce( lambda x,y: x+y, z)
elif options.group_function == "mean":
f = scipy.mean
elif options.group_function == "cat":
f = lambda x: ";".join( [ y for y in x if y != "" ] )
converter = str
elif options.group_function == "uniq":
f = lambda x: ";".join( [ y for y in set(x) if y != "" ] )
converter = str
elif options.group_function == "stats":
f = lambda x: str(Stats.DistributionalParameters(x))
# update headers
new_fields = [ fields[options.group_column] ]
for c in options.columns:
new_fields += list( map(lambda x: "%s_%s" % (fields[c], x), Stats.DistributionalParameters().getHeaders() ) )
## convert values to floats (except for group_column)
## Delete rows with unconvertable values and not in options.columns
new_table = []
for row in table:
skip = False
new_row = [ row[options.group_column] ]
for c in options.columns:
if row[c] == options.missing_value:
new_row.append(row[c])
else:
try:
new_row.append( converter(row[c]) )
except ValueError:
skip = True
break
if not skip: new_table.append(new_row)
table = new_table
new_rows = CSV.GroupTable( table,
group_column = 0,
group_function = f )
options.stdout.write("\t".join(new_fields) + "\n")
for row in new_rows:
options.stdout.write( "\t".join( map(str,row) ) + "\n")
