exposed the list of values to be considered as null in the Cleaner

Author: Lisa McBride

skrub/_clean_null_strings.py

@@ -231,9 +231,19 @@ class CleanNullStrings(SingleColumnTransformer):
         ...
     skrub._single_column_transformer.RejectColumn: Column 's' does not contain strings.
     """
+    def __init__(
+        self,
+        null_strings=None,
+    ):
+        self.null_strings = null_strings
 
     def fit_transform(self, column, y=None):
         del y
+        self.null_list_ = STR_NA_VALUES
+        if self.null_strings is not None:
+            if not (isinstance(self.null_strings, str) or (isinstance(self.null_strings, list) and all(isinstance(i, str) for i in self.null_strings))):
+                raise TypeError("Expected either a string or a list of strictly strings.")
+            self.null_list_ += self.null_strings
         if not (sbd.is_pandas_object(column) or sbd.is_string(column)):
             raise RejectColumn(f"Column {sbd.name(column)!r} does not contain strings.")
         return self.transform(column)
@@ -242,5 +252,5 @@ def transform(self, column):
         if not (sbd.is_pandas_object(column) or sbd.is_string(column)):
             return column
         column = _trim_whitespace_only(column)
-        column = sbd.replace(column, STR_NA_VALUES, sbd.null_value_for(column))
+        column = sbd.replace(column, self.null_list_, sbd.null_value_for(column))
         return column

skrub/_table_vectorizer.py

@@ -115,6 +115,7 @@ def _check_transformer(transformer):
 def _get_preprocessors(
     *,
     cols,
+    null_strings,
     drop_null_fraction,
     drop_if_unique,
     drop_if_constant,
@@ -125,7 +126,9 @@ def _get_preprocessors(
 ):
     steps = [CheckInputDataFrame()]
     transformers = [
-        CleanNullStrings(),
+        CleanNullStrings(null_strings=null_strings,
+
+        ),
         DropUninformative(
             drop_null_fraction=drop_null_fraction,
             drop_if_constant=drop_if_constant,
@@ -321,6 +324,7 @@ class Cleaner(TransformerMixin, BaseEstimator):
 
     def __init__(
         self,
+        null_strings=None,
         drop_null_fraction=1.0,
         drop_if_constant=False,
         drop_if_unique=False,
@@ -329,6 +333,7 @@ def __init__(
         cast_to_str=False,
         n_jobs=1,
     ):
+        self.null_strings = null_strings
         self.drop_null_fraction = drop_null_fraction
         self.drop_if_constant = drop_if_constant
         self.drop_if_unique = drop_if_unique
@@ -365,6 +370,7 @@ def fit_transform(self, X, y=None):
 
         all_steps = _get_preprocessors(
             cols=s.all(),
+            null_strings=self.null_strings,
             drop_null_fraction=self.drop_null_fraction,
             drop_if_constant=self.drop_if_constant,
             drop_if_unique=self.drop_if_unique,