FEAT: Include only referenced data columns in chart specs

altair/utils/data.py

@@ -21,7 +21,9 @@
 
 
 class DataTransformerRegistry(PluginRegistry[DataTransformerType]):
-    _global_settings = {"consolidate_datasets": True}
+    _global_settings = {
+        "consolidate_datasets": True,
+    }
 
     @property
     def consolidate_datasets(self):
@@ -31,6 +33,40 @@ def consolidate_datasets(self):
     def consolidate_datasets(self, value):
         self._global_settings["consolidate_datasets"] = value
 
+    def compress_datasets(self, enable=True, transformed_charts=False):
+        """Whether to remove data fields that do not occurr elsewhere in the chart spec
+
+        The approach taken here is conservative as it will
+        leave some of the data fields that should be removed because they
+        have a common or short name that happens to occur elsewhere in the
+        spec by coincidence.
+
+        Parameters
+        ----------
+
+        enable : bool
+            Whether compression is enabled.
+
+        transformed_charts : bool
+            Whether to enable compression for transformed charts.
+            Data compression can break charts with transforms where data
+            fields are referenced without the field name occurring
+            literally in the chart spec.
+        """
+        if transformed_charts:
+            warnings.warn(
+                "Data compression can break charts with transforms where data"
+                " fields are referenced without the field name occurring"
+                " literally in the chart spec. Please revert this option if you"
+                " are observing unexpected behavior with transformed charts."
+            )
+        self._compress_datasets = enable
+        self._compress_transformed_charts = transformed_charts
+        return
+
+    _compress_datasets = True
+    _compress_transformed_charts = False
+
 
 # ==============================================================================
 # Data model transformers

altair/vegalite/v5/api.py

@@ -103,6 +103,34 @@ def _prepare_data(data, context=None):
     return data
 
 
+def _compress_datasets(dct):
+    """Remove data fields that do not occurr elsewhere in the chart spec
+
+    The approach taken here is conservative as it will
+    leave some of the data fields that should be removed because they
+    have a common or short name that happens to occur elsewhere in the
+    spec by coincidence.
+    It might incorreclty remove fields when advanced transforms are used.
+    Removing data fields properly is difficult as mentioned here
+    https://github.com/altair-viz/altair/issues/2428#issuecomment-798917267.
+
+    Parameters
+    ----------
+    dct : dict
+        The input chart schema
+    """
+    spec_as_string = [
+        val for key, val in dct.items() if key not in ["data", "datasets", "$schema"]
+    ].__str__()
+    for dataset_name in dct["datasets"]:
+        for dict_row in dct["datasets"][dataset_name]:
+            # Must iterate over a list since the dict will change size
+            for field in list(dict_row.keys()):
+                if field not in spec_as_string:
+                    del dict_row[field]
+    return dct
+
+
 # ------------------------------------------------------------------------
 # Aliases & specializations
 Bin = core.BinParams
@@ -568,6 +596,14 @@ def to_dict(self, *args, **kwargs):
             if context["datasets"]:
                 dct.setdefault("datasets", {}).update(context["datasets"])
 
+        # Remove data fields that do not occurr elsewhere in the chart spec
+        if data_transformers._compress_datasets:
+            if (
+                not data_transformers._compress_transformed_charts
+                and "transform" not in dct.keys()
+            ) or data_transformers._compress_transformed_charts:
+                if "datasets" in dct:
+                    dct = _compress_datasets(dct)
         return dct
 
     def to_html(