Convert only the columns used in the query

In the end, it's only worth calling to_numeric on the columns used for
numerical comparison.

This gets us halfway there, since in most cases, only a small subset of
metadata columns are used in the query.

This is a hacky approach, but it is more computationally efficient.

augur/filter/include_exclude_rules.py

@@ -13,6 +13,12 @@
 from augur.utils import read_strains
 from . import constants
 
+try:
+    # python ≥3.8 only
+    from typing import Literal  # type: ignore
+except ImportError:
+    from typing_extensions import Literal  # type: ignore
+
 
 # The strains to keep as a result of applying a filter function.
 FilterFunctionReturn = Set[str]
@@ -178,8 +184,8 @@ def filter_by_query(metadata, query) -> FilterFunctionReturn:
     set()
 
     """
-    # Try converting all columns to numeric.
-    for column in metadata.columns:
+    # Try converting all queried columns to numeric.
+    for column in extract_variables(query).intersection(metadata.columns):
         metadata[column] = pd.to_numeric(metadata[column], errors='ignore')
 
     return set(metadata.query(query).index.values)
@@ -803,3 +809,29 @@ def _filter_kwargs_to_str(kwargs: FilterFunctionKwargs):
         kwarg_list.append((key, value))
 
     return json.dumps(kwarg_list)
+
+
+# From https://stackoverflow.com/a/76536356
+def extract_variables(pandas_query: str):
+    """Extract variable names used in a pandas query string."""
+
+    # Track variables in a dictionary to be used as a dictionary of globals.
+    variables: Dict[str, Literal[None]] = {}
+
+    while True:
+        try:
+            # Try creating a Expr object with the query string and dictionary of globals.
+            # This will raise an error as long as the dictionary of globals is incomplete.
+            env = pd.core.computation.scope.ensure_scope(level=0, global_dict=variables)
+            pd.core.computation.expr.Expr(pandas_query, env=env)
+
+            # Exit the loop when evaluation is successful.
+            break
+        except pd.errors.UndefinedVariableError as e:
+            # This relies on the format defined here: https://github.com/pandas-dev/pandas/blob/965ceca9fd796940050d6fc817707bba1c4f9bff/pandas/errors/__init__.py#L401
+            name = re.findall("name '(.+?)' is not defined", str(e))[0]
+
+            # Add the name to the globals dictionary with a dummy value.
+            variables[name] = None
+
+    return set(variables.keys())

setup.py

@@ -87,6 +87,7 @@
             "sphinx-autodoc-typehints >=1.21.4",
             "types-jsonschema >=3.0.0, ==3.*",
             "types-setuptools",
+            "typing_extensions; python_version <'3.8'",
             "wheel >=0.32.3",
             "ipdb >=0.10.1"
         ]