def applyMaskString(self, maskString, bAdd=True):
df = self.data
Index = np.array(range(df.shape[0]))
sMask = maskString.replace('{Index}', 'Index')
for i, c in enumerate(self.columns):
c_no_unit = no_unit(c).strip()
c_in_df = df.columns[i]
# TODO sort out the mess with asarray (introduced to have and/or
# as array won't work with date comparison
# NOTE: using iloc to avoid duplicates column issue
if isinstance(df.iloc[0, i], pd._libs.tslibs.timestamps.Timestamp):
sMask = sMask.replace('{' + c_no_unit + '}',
'df[\'' + c_in_df + '\']')
else:
sMask = sMask.replace('{' + c_no_unit + '}',
'np.asarray(df[\'' + c_in_df + '\'])')
df_new = None
name_new = None
if len(sMask.strip()) > 0 and sMask.strip().lower() != 'no mask':
try:
mask = np.asarray(eval(sMask))
if bAdd:
df_new = df[mask]
name_new = self.raw_name + '_masked'
else:
self.mask = mask
self.maskString = maskString
except:
raise Exception('Error: The mask failed for table: ' +
self.name)
return df_new, name_new
def evalFormula(self,sFormula):
df = self.data
Index = np.array(range(df.shape[0]))
sFormula=sFormula.replace('{Index}','Index')
for i,c in enumerate(self.columns):
c_no_unit = no_unit(c).strip()
c_in_df = df.columns[i]
sFormula=sFormula.replace('{'+c_no_unit+'}','df[\''+c_in_df+'\']')
#print(sFormula)
try:
NewCol=eval(sFormula)
return NewCol
except:
return None
def columns_clean(self):
return [no_unit(s) for s in self.columns]