diff --git a/skrub/__init__.py b/skrub/__init__.py
index 67143ec9a..f4564e106 100644
--- a/skrub/__init__.py
+++ b/skrub/__init__.py
@@ -41,6 +41,7 @@
 from ._similarity_encoder import SimilarityEncoder
 from ._squashing_scaler import SquashingScaler
 from ._string_encoder import StringEncoder
+from ._string_parser import StringParser
 from ._table_vectorizer import Cleaner, TableVectorizer
 from ._tabular_pipeline import tabular_learner, tabular_pipeline
 from ._text_encoder import TextEncoder
@@ -104,4 +105,5 @@
     "ApplyToCols",
     "ApplyToFrame",
     "ToFloat",
+    "StringParser",
 ]
diff --git a/skrub/_string_parser.py b/skrub/_string_parser.py
new file mode 100644
index 000000000..6083405cf
--- /dev/null
+++ b/skrub/_string_parser.py
@@ -0,0 +1,51 @@
+from . import _dataframe as sbd
+from ._on_each_column import SingleColumnTransformer
+
+
+class StringParser(SingleColumnTransformer):
+    """
+    """
+    def __init__(self, dictionary):
+        self.dictionary = dictionary
+    
+    def fit_transform(self, X, y=None):
+        del y
+
+        return self.transform(X)
+
+    def transform(self, X):
+        breakpoint()
+        col_name = sbd.name(X) or "string"
+
+        X_str = X.astype(str).str.lower().str.strip()
+
+        pattern = (
+            r"(?:(?P<unit1>[a-zA-Z]+)\s*(?P<value1>[-+]?\d*\.?)"
+            r"|(?:(?P<value2>[-+]?\d*\.?\d+)\s*(?P<unit2>[a-zA-Z]+))")
+        
+        parsed = X_str.str.extract(pattern)
+        parsed["unit"] = parsed["unit1"].combine_first(parsed["unit2"])
+        parsed["value"] = parsed["value1"].combine_first(parsed["value2"])
+        
+        parsed = parsed.drop(columns=["unit1", "value1", "unit2", "value2"])
+
+        parsed["value"] = parsed["value"].str.replace(",", ".").astype(float)
+
+        base_unit = next(
+            (u for u, f in self.dictionary.items() if f == 1.0),
+            list(self.dictionary.keys())[0],
+        )
+
+        parsed["unit"] = parsed["unit"].fillna(base_unit)
+
+        parsed["factor"] = parsed["unit"].map(self.dictionary)
+        result = parsed["value"] * parsed["factor"]
+        result.name = f"{col_name}_{base_unit}"
+        result.index = X.index
+
+        self.input_name_ = col_name
+        self.all_outputs_ = [result.name]
+        self._is_fitted = True
+
+        return result
+