def handle_missing(self, fltr):
if self.cfg is None or not self.cfg.get("missing", False):
return fltr
return {
"missing": True,
"query": "{col} != {col}".format(col=build_col_key(self.column)),
}
def build_filter(self):
if self.cfg is None:
return super(DateFilter, self).handle_missing(None)
start, end = (self.cfg.get(p) for p in ["start", "end"])
fltr = dict(start=start, end=end)
queries = []
if start:
queries.append("{} >= '{}'".format(build_col_key(self.column),
start))
if end:
queries.append("{} <= '{}'".format(build_col_key(self.column),
end))
if len(queries) == 2 and start == end:
queries = ["{} == '{}'".format(build_col_key(self.column), start)]
if not len(queries):
return super(DateFilter, self).handle_missing(None)
fltr["query"] = " and ".join(queries)
return super(DateFilter, self).handle_missing(fltr)
def build_filter(self):
if self.cfg is None:
return super(NumericFilter, self).handle_missing(None)
cfg_val, cfg_operand, cfg_min, cfg_max = (
self.cfg.get(p) for p in ["value", "operand", "min", "max"])
if cfg_operand in ["=", "ne"]:
state = make_list(cfg_val or [])
if not len(state):
return super(NumericFilter, self).handle_missing(None)
fltr = dict(value=cfg_val, operand=cfg_operand)
if len(state) == 1:
fltr["query"] = "{} {} {}".format(
build_col_key(self.column),
"==" if cfg_operand == "=" else "!=",
state[0],
)
else:
fltr["query"] = "{} {} ({})".format(
build_col_key(self.column),
"in" if cfg_operand == "=" else "not in",
", ".join(map(str, state)),
)
return super(NumericFilter, self).handle_missing(fltr)
if cfg_operand in ["<", ">", "<=", ">="]:
if cfg_val is None:
return super(NumericFilter, self).handle_missing(None)
fltr = dict(
value=cfg_val,
operand=cfg_operand,
query="{} {} {}".format(build_col_key(self.column),
cfg_operand, cfg_val),
)
return super(NumericFilter, self).handle_missing(fltr)
if cfg_operand in ["[]", "()"]:
fltr = dict(operand=cfg_operand)
queries = []
if cfg_min is not None:
fltr["min"] = cfg_min
queries.append("{} >{} {}".format(
build_col_key(self.column),
"=" if cfg_operand == "[]" else "",
cfg_min,
))
if cfg_max is not None:
fltr["max"] = cfg_max
queries.append("{} <{} {}".format(
build_col_key(self.column),
"=" if cfg_operand == "[]" else "",
cfg_max,
))
if len(queries) == 2 and cfg_max == cfg_min:
queries = [
"{} == {}".format(build_col_key(self.column), cfg_max)
]
if not len(queries):
return super(NumericFilter, self).handle_missing(None)
fltr["query"] = " and ".join(queries)
return super(NumericFilter, self).handle_missing(fltr)
return super(NumericFilter, self).handle_missing(None)
def build_filter(self):
if self.cfg is None or not len(self.cfg.get("value", [])):
return super(StringFilter, self).handle_missing(None)
state = self.cfg.get("value", [])
operand = self.cfg.get("operand", "=")
fltr = dict(value=state)
if len(state) == 1:
val_str = ("'{}'" if self.classification == "S" else "{}").format(
state[0])
fltr["query"] = "{} {} {}".format(build_col_key(self.column),
"==" if operand == "=" else "!=",
val_str)
else:
val_str = ("'{}'".format("', '".join(state))
if self.classification == "S" else ",".join(state))
fltr["query"] = "{} {} ({})".format(
build_col_key(self.column),
"in" if operand == "=" else "not in",
val_str,
)
return super(StringFilter, self).handle_missing(fltr)
def group_filter_handler(col_def, group_val, group_classifier):
col_def_segs = col_def.split("|")
if len(col_def_segs) > 1:
col, freq = col_def_segs
if group_val == "nan":
return "{col} != {col}".format(
col=build_col_key(col)), "{}: NaN".format(col)
if freq == "WD":
return (
"{}.dt.dayofweek == {}".format(build_col_key(col), group_val),
"{}.dt.dayofweek: {}".format(col, group_val),
)
elif freq == "H2":
return (
"{}.dt.hour == {}".format(build_col_key(col), group_val),
"{}.dt.hour: {}".format(col, group_val),
)
elif freq == "H":
ts_val = pd.Timestamp(group_val)
day = ts_val.strftime("%Y%m%d")
hour = ts_val.hour
return (
"{col}.dt.date == '{day}' and {col}.dt.hour == {hour}".format(
col=build_col_key(col), day=day, hour=hour),
"{col}.dt.date: {day}, {col}.dt.hour: {hour}".format(
col=col, day=day, hour=hour),
)
elif freq == "D":
ts_val = convert_date_val_to_date(group_val)
day = ts_val.strftime("%Y%m%d")
return (
"{col}.dt.date == '{day}'".format(col=build_col_key(col),
day=day),
"{}.dt.date: {}".format(col, day),
)
elif freq == "W":
ts_val = convert_date_val_to_date(group_val)
return (
"{col}.dt.year == {year} and {col}.dt.week == {week}".format(
col=build_col_key(col), year=ts_val.year,
week=ts_val.week),
"{col}.dt.year: {year}, {col}.dt.week: {week}".format(
col=col, year=ts_val.year, week=ts_val.week),
)
elif freq == "M":
ts_val = convert_date_val_to_date(group_val)
return (
"{col}.dt.year == {year} and {col}.dt.month == {month}".format(
col=build_col_key(col),
year=ts_val.year,
month=ts_val.month),
"{col}.dt.year: {year}, {col}.dt.month: {month}".format(
col=col, year=ts_val.year, month=ts_val.month),
)
elif freq == "Q":
ts_val = convert_date_val_to_date(group_val)
return (
"{col}.dt.year == {year} and {col}.dt.quarter == {quarter}".
format(col=build_col_key(col),
year=ts_val.year,
quarter=ts_val.quarter),
"{col}.dt.year: {year}, {col}.dt.quarter: {quarter}".format(
col=col, year=ts_val.year, quarter=ts_val.quarter),
)
elif freq == "Y":
ts_val = convert_date_val_to_date(group_val)
return (
"{col}.dt.year == {year}".format(col=build_col_key(col),
year=ts_val.year),
"{}.dt.year: {}".format(col, ts_val.year),
)
if group_val == "nan":
return "{col} != {col}".format(
col=build_col_key(col_def)), "{}: NaN".format(col_def)
if group_classifier in ["I", "F", "B"]:
return (
"{col} == {val}".format(col=build_col_key(col_def), val=group_val),
"{}: {}".format(col_def, group_val),
)
if group_classifier == "D":
group_val = convert_date_val_to_date(group_val).strftime("%Y%m%d")
return (
"{col} == '{val}'".format(col=build_col_key(col_def), val=group_val),
"{}: {}".format(col_def, group_val),
)
def build_filter(self):
if self.cfg is None:
return super(StringFilter, self).handle_missing(None)
action = self.cfg.get("action", "equals")
if action == "equals" and not len(self.cfg.get("value", [])):
return super(StringFilter, self).handle_missing(None)
elif action != "equals" and not self.cfg.get("raw"):
return super(StringFilter, self).handle_missing(None)
state = self.cfg.get("value", [])
case_sensitive = self.cfg.get("caseSensitive", False)
operand = self.cfg.get("operand", "=")
raw = self.cfg.get("raw")
fltr = dict(
value=state,
operand=operand,
caseSensitive=case_sensitive,
action=action,
raw=raw,
)
if action == "equals":
if len(state) == 1:
val_str = ("'{}'" if self.classification == "S" else
"{}").format(state[0])
fltr["query"] = "{} {} {}".format(
build_col_key(self.column),
"==" if operand == "=" else "!=",
val_str,
)
else:
val_str = ("'{}'".format("', '".join(state))
if self.classification == "S" else ",".join(state))
fltr["query"] = "{} {} ({})".format(
build_col_key(self.column),
"in" if operand == "=" else "not in",
val_str,
)
elif action in ["startswith", "endswith"]:
case_insensitive_conversion = "" if case_sensitive else ".str.lower()"
fltr["query"] = "{}{}.str.{}('{}', na=False)".format(
build_col_key(self.column),
case_insensitive_conversion,
action,
raw if case_sensitive else raw.lower(),
)
fltr["query"] = handle_ne(fltr["query"], operand)
elif action == "contains":
fltr["query"] = "{}.str.contains('{}', na=False, case={})".format(
build_col_key(self.column),
raw,
"True" if case_sensitive else "False",
)
fltr["query"] = handle_ne(fltr["query"], operand)
elif action == "length":
if "," in raw:
start, end = raw.split(",")
fltr["query"] = "{start} <= {col}.str.len() <= {end}".format(
col=build_col_key(self.column),
start=start,
end=end,
)
else:
fltr["query"] = "{}.str.len() == {}".format(
build_col_key(self.column), raw)
fltr["query"] = handle_ne(fltr["query"], operand)
return super(StringFilter, self).handle_missing(fltr)
