flag columns are now created if a variable is actually touched

core.py

@@ -39,11 +39,12 @@ def flagNext(flagger: BaseFlagger, flags: pd.Series, n: int) -> pd.Series:
 def runner(meta, flagger, data, flags=None, nodata=np.nan):
 
     if flags is None:
+        # flags = pd.DataFrame(index=data.index)
         flags = flagger.emptyFlags(data)
-    else:
-        if not all(flags.columns == flagger.emptyFlags(data.iloc[0]).columns):
-            raise TypeError("structure of given flag does not "
-                            "correspond to flagger requirements")
+    # else:
+    #     if not all(flags.columns == flagger.emptyFlags(data.iloc[0]).columns):
+    #         raise TypeError("structure of given flag does not "
+    #                         "correspond to flagger requirements")
 
     # NOTE:
     # We need an index frequency in order to calculate ticks
@@ -77,9 +78,14 @@ def runner(meta, flagger, data, flags=None, nodata=np.nan):
             varname, start_date, end_date = configrow[fields]
             func_name, flag_params = parseFlag(flag_test)
 
-            if flag_params.get(FlagParams.ASSIGN):
+            # NOTE:
+            # create a flag column if this is explicitly stated
+            # or if a variable is checked but no corresponding
+            # flag column exists
+            if (flag_params.get(FlagParams.ASSIGN) or
+                ((varname in data) and (varname not in flags))):
                 dummy = pd.DataFrame(index=data.index, columns=[varname])
-                flags[varname] = flagger.emptyFlags(dummy)
+                flags = flags.join(flagger.initFlags(dummy))
 
             if varname not in data:
                 continue
@@ -104,18 +110,20 @@ def runner(meta, flagger, data, flags=None, nodata=np.nan):
                 raise NameError(
                     f"function name {func_name} is not definied (variable '{varname}, 'line: {idx + 1})")
 
-
             # flag a timespan after the condition is met,
             # duration given in 'flag_period'
             flag_period = flag_params.pop(Params.FLAGPERIOD, None)
             if flag_period:
-                flag_params[Params.FLAGVALUES] = _periodToTicks(flag_period, data.index.freq)
+                flag_params[Params.FLAGVALUES] = _periodToTicks(
+                    flag_period, data.index.freq)
 
             # flag a certain amount of values after condition is met,
             # number given in 'flag_values'
             flag_values = flag_params.pop(Params.FLAGVALUES, None)
             if flag_values:
-                fchunk[varname] = flagNext(flagger, fchunk[varname], flag_values)
+                fchunk[varname] = flagNext(flagger,
+                                           fchunk[varname],
+                                           flag_values)
 
             data.loc[start_date:end_date] = dchunk
             flags.loc[start_date:end_date] = fchunk

flagger/baseflagger.py

@@ -30,13 +30,16 @@ class BaseFlagger:
         flags[:] = flag
         return flags.values
 
-    def emptyFlags(self,
-                   data: pd.DataFrame,
-                   value: Optional[Number] = np.nan) -> pd.DataFrame:
+    def initFlags(self,
+                  data: pd.DataFrame,
+                  value: Optional[Number] = np.nan) -> pd.DataFrame:
         out = data.copy()
         out[:] = value
         return out
 
+    def emptyFlags(self, data: pd.DataFrame) -> pd.DataFrame:
+        return pd.DataFrame(index=data.index)
+
     def isFlagged(self, flags: ArrayLike, flag: T = None) -> ArrayLike:
         if flag is None:
             return (pd.notnull(flags) & (flags != self.no_flag))

flagger/dmpflagger.py

@@ -31,14 +31,27 @@ class DmpFlagger(BaseFlagger):
         super().__init__(no_flag, flag)
         self.flag_fields = [FlagFields.FLAG, FlagFields.CAUSE, FlagFields.COMMENT]
 
-    def emptyFlags(self, data, value="NIL", **kwargs):
+    def emptyFlags(self, data, **kwargs):
+        columns = pd.MultiIndex(
+            levels=[[], []],
+            labels=[[], []],
+            names=[ColumnLevels.VARIABLES, ColumnLevels.FLAGS])
+        return pd.DataFrame(index=data.index, columns=columns)
+
+    # def _getColumns(self, data):
+    #     if isinstance(data, pd.DataFrame):
+    #         return data.columns
+    #     return [data.name]
+
+    def initFlags(self, data, value="NIL", **kwargs):
         columns = data.columns if isinstance(data, pd.DataFrame) else [data.name]
         columns = pd.MultiIndex.from_product(
             [columns, self.flag_fields],
             names=[ColumnLevels.VARIABLES, ColumnLevels.FLAGS])
         return pd.DataFrame(data=value, columns=columns, index=data.index)
 
-    def setFlag(self, flags, flag=Flags.BAD, cause="NIL", comment="NIL", **kwargs):
+    def setFlag(self, flags, flag=Flags.BAD,
+                cause="NIL", comment="NIL", **kwargs):
         self._isFlag(flag)
         flags = self._reduceColumns(flags)
         for field, f in zip(self.flag_fields, [flag, cause, comment]):