def perf_func(col_name, entries, col_funcs, grouping=None):
'''
Given a dictionary of column functions mapping to column name, perform
column function on column entries. Do it by groups if there is grouping
:param col_name: string column name
:param entries: a list of values
:param col_funcs: a dict {col_name: func_name}
:param grouping: a list of [indexes], each sublist is a equivalence class
:return: a list of entries [val1, val2...]
'''
if grouping:
# there are grouping
if col_funcs.get(col_name):
func_name = col_funcs[col_name]
# Perform column function on each grouping of the entries
perf_entries = [eva.FUNC[func_name](man.get_entries(entries, grp))
for grp in grouping]
else:
# No column function, return the first element of each group
perf_entries = [entries[grp[0]] for grp in grouping]
else:
# There are no grouping
if col_funcs.get(col_name):
# There is column function for the column
# Get the name of the column function
func_name = col_funcs[col_name]
# perform col functions on column entries
perf_entries = [eva.FUNC[func_name](entries)]
else:
# if there's no need to perform col func, return entries as it is
perf_entries = entries
return perf_entries
def filter_grps(line, tb, grouping):
'''
Filter out groups that do not meet filter condition before performing
column functions on groups in do_col_func_two
:param line: a TENIENDO condition without ' TENIENDO ' and '()'
:param tb: a dict {col_name: [entries]}
:param grouping: a list of [indexes], each sublist is an equivalence class
:return: a list of [indexes], each equivalence class meets the condition
'''
# split the line into var1, operator, var2
var1, oper, var2 = split_clauses(line)
if '(' in var1:
col_str = var1
val_str = var2
else:
col_str = var2
val_str = var1
# Get the list of function names
func_keys = list(eva.FUNC.keys())
col, func = [(strip_func(col_str), func) for func in func_keys
if func + '(' in col_str][0]
if tb.get(col) is None:
raise hpr.InvalidQueryError('Table do not have column: {0}'.format(col))
# Get the list of entry values of each group after performing function
values = [eva.FUNC[func](man.get_entries(tb[col], grp)) for grp in grouping]
# Filter list of values by the var2 using operator, return a list of group
# ids of groups that passed filter condition
if col_str == var1:
filtered_grps = [grp_id for grp_id, val in enumerate(values)
if eva.OPER[oper](var1=val,
var2=hpr.parse_val(val_str))]
else:
filtered_grps = [grp_id for grp_id, val in enumerate(values)
if eva.OPER[oper](var1=hpr.parse_val(val_str),
var2=val)]
return [grouping[i] for i in filtered_grps]
