def eval_separate(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
self.assertArg(node, args,
index=1,
cond=lambda x: x <= n_cols,
capture_indices=[0])
tbl = robjects.r(args[0])
col = tbl.columns[int(args[1]) - 1]
ret_df_name = get_fresh_name()
_script = '{ret_df} <- separate({table}, {col1}, c("{TMP1}", "{TMP2}"))'.format(
ret_df=ret_df_name, table=args[0], col1=str(args[1]), TMP1=get_fresh_col(), TMP2=get_fresh_col())
if tbl[col].dtype == np.object:
cell = tbl[col][0]
if cell.count('_') > 1:
_script = '{ret_df} <- separate({table}, {col1}, c("{TMP1}", "{TMP2}", "{TMP3}"), sep="_")'.format(
ret_df=ret_df_name, table=args[0], col1=str(args[1]), TMP1=get_fresh_col(), TMP2=get_fresh_col(), TMP3=get_fresh_col())
else:
raise GeneralError()
try:
ret_val = robjects.r(_script)
return ret_df_name
except:
logger.error('Error in interpreting separate...')
raise GeneralError()
def eval_mutate(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
self.assertArg(node, args,
index=2,
cond=lambda x: x <= n_cols,
capture_indices=[0])
self.assertArg(node, args,
index=3,
cond=lambda x: x <= n_cols,
capture_indices=[0])
self.assertArg(node, args,
index=2,
cond=lambda x: get_type(args[0], str(x)) == 'numeric',
capture_indices=[0])
self.assertArg(node, args,
index=3,
cond=lambda x: get_type(args[0], str(x)) == 'numeric',
capture_indices=[0])
ret_df_name = get_fresh_name()
_script = '{ret_df} <- {table} %>% mutate({TMP}=.[[{col1}]] {op} .[[{col2}]])'.format(
ret_df=ret_df_name, table=args[0], TMP='mutate_a', op=args[1], col1=str(args[2]), col2=str(args[3]))
# _script = '{ret_df} <- {table} %>% mutate({TMP}=.[[{col1}]] {op} .[[{col2}]])'.format(
# ret_df=ret_df_name, table=args[0], TMP=get_fresh_col(), op=args[1], col1=str(args[2]), col2=str(args[3]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except:
logger.error('Error in interpreting mutate...')
raise GeneralError()
def eval_summarise(self, node, args):
input_tbl = robjects.r(args[0])
input_cols = input_tbl.columns.values
n_cols = len(input_cols)
aggr_fun = str(args[1])
self.assertArg(node, args,
index=2,
cond=lambda x: x <= n_cols,
capture_indices=[0])
if not aggr_fun == 'n':
self.assertArg(node, args,
index=2,
cond=lambda x: get_type(args[0], str(x)) == 'integer' or get_type(args[0], str(x)) == 'numeric',
capture_indices=[0])
ret_df_name = get_fresh_name()
_script = ''
if aggr_fun == 'n':
_script = '{ret_df} <- {table} %>% summarise({TMP} = {aggr} ())'.format(
ret_df=ret_df_name, table=args[0], TMP=get_fresh_col(), aggr=aggr_fun)
else:
aggr_col = input_cols[args[2]-1]
_script = '{ret_df} <- {table} %>% summarise({TMP} = {aggr} (`{col}`))'.format(
ret_df=ret_df_name, table=args[0], TMP=get_fresh_col(), aggr=aggr_fun, col=aggr_col)
try:
ret_val = robjects.r(_script)
return ret_df_name
except Exception as e:
logger.error('Error in interpreting summarise...')
raise GeneralError()
def eval_spread(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
first_idx = int(args[1])
self.assertArg(node, args,
index=1,
cond=lambda x: x <= n_cols,
capture_indices=[0])
self.assertArg(node, args,
index=2,
cond=lambda x: x <= n_cols and x > first_idx,
capture_indices=[0, 1])
ret_df_name = get_fresh_name()
_script = '{ret_df} <- spread({table}, {col1}, {col2})'.format(
ret_df=ret_df_name, table=args[0], col1=str(args[1]), col2=str(args[2]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except:
logger.error('Error in interpreting spread...')
# r0 = robjects.r(args[0])
# logger.info(r0)
# temp_df_name = get_fresh_name()
# key_script = '{ret_df} <- select({table}, {cols})'.format(
# ret_df=ret_df_name, table=args[0], cols=get_collist([str(args[1])]))
# rv = robjects.r(key_script)
# temp_df_name = get_fresh_name()
# id_script = '{ret_df} <- select({table}, {cols})'.format(
# ret_df=ret_df_name, table=args[0], cols=get_collist(["-"+str(args[1]), "-"+str(args[2])]))
# rv2 = robjects.r(id_script)
raise GeneralError()
def eval_inner_join(self, node, args):
ret_df_name = get_fresh_name()
_script = '{ret_df} <- inner_join({t1}, {t2})'.format(
ret_df=ret_df_name, t1=args[0], t2=args[1])
try:
ret_val = robjects.r(_script)
return ret_df_name
except:
logger.error('Error in interpreting innerjoin...')
raise GeneralError()
def eval_mutateCustom(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
col_idx = args[2] - 1
self.assertArg(node, args,
index=2,
cond=lambda x: x <= n_cols,
capture_indices=[0])
input_tbl = robjects.r(args[0])
col_type = input_tbl.dtypes[col_idx]
if col_type == np.float64 or col_type == np.int64:
raise GeneralError()
ret_df_name = get_fresh_name()
_script = '{ret_df} <- {table} %>% mutate({TMP}=(.[[{col1}]] {op} "{col2}"))'.format(
ret_df=ret_df_name, table=args[0], TMP=get_fresh_col(), op=args[1], col1=str(args[2]), col2=str(args[3]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except Exception as e:
logger.error('Error in interpreting mutateCustom...', _script)
# assert False, e
raise GeneralError()
def eval_cumsum(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
self.assertArg(node, args,
index=1,
cond=lambda x: x <= n_cols,
capture_indices=[0])
ret_df_name = get_fresh_name()
_script = '{ret_df} <- {table} %>% mutate({TMP}=cumsum(.[[{col1}]]))'.format(
ret_df=ret_df_name, table=args[0], TMP='cumsum', col1=str(args[1]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except Exception as e:
logger.error('Error in interpreting cumsum...', _script)
raise GeneralError()
def eval_gather(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
self.assertArg(node, args,
index=1,
cond=lambda x: max(list(map(lambda y: int(y), x))) <= n_cols,
capture_indices=[0])
ret_df_name = get_fresh_name()
_script = '{ret_df} <- gather({table}, KEY, VALUE, {cols})'.format(
ret_df=ret_df_name, table=args[0], cols=get_collist(args[1]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except:
logger.error('Error in interpreting gather...')
raise GeneralError()
def eval_filter(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
self.assertArg(node, args,
index=2,
cond=lambda x: x <= n_cols,
capture_indices=[0])
self.assertArg(node, args,
index=2,
cond=lambda x: get_type(args[0], str(x)) != 'factor',
capture_indices=[0])
ret_df_name = get_fresh_name()
_script = '{ret_df} <- {table} %>% filter(.[[{col}]] {op} "{const}")'.format(
ret_df=ret_df_name, table=args[0], op=args[1], col=str(args[2]), const=str(args[3]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except Exception as e:
logger.error('Error in interpreting filter...', e)
raise GeneralError()
def eval_unite(self, node, args):
n_cols = robjects.r('ncol(' + args[0] + ')')[0]
first_idx = int(args[1])
self.assertArg(node, args,
index=1,
cond=lambda x: x <= n_cols,
capture_indices=[0])
self.assertArg(node, args,
index=2,
cond=lambda x: x <= n_cols and x != first_idx,
capture_indices=[0, 1])
ret_df_name = get_fresh_name()
_script = '{ret_df} <- unite({table}, {TMP}, {col1}, {col2})'.format(
ret_df=ret_df_name, table=args[0], TMP=get_fresh_col(), col1=str(args[1]), col2=str(args[2]))
try:
ret_val = robjects.r(_script)
return ret_df_name
except:
logger.error('Error in interpreting unite...')
raise GeneralError()