def accessed_columns(expr):
"""Return a set of column indexes accessed by an expression.
Assumes that named attribute references have been converted to integer
positions.
"""
for ex in expr.walk():
assert not isinstance(ex, NamedAttributeRef)
return set([ex.position for ex in expr.walk()
if isinstance(ex, UnnamedAttributeRef)])
def udf_undefined_vars(expr, vars):
"""Return a list of undefined variables in a UDF.
:param expr: An expression corresponding to a UDF. Variable references are
identified by instances of NamedAttributeRef.
:param vars: A list of variables in the argument list to the function.
:type vars: list of strings
"""
return [ex.name for ex in expr.walk()
if isinstance(ex, NamedAttributeRef) and ex.name not in vars]
def reindex_expr(expr, index_map):
"""Changes references to key columns to references to value columns in
index_map.
Assumes that named attribute references have been converted to integer
positions.
"""
for ex in expr.walk():
assert (not isinstance(ex, AttributeRef)
or isinstance(ex, UnnamedAttributeRef))
if isinstance(ex, UnnamedAttributeRef) and ex.position in index_map:
ex.position = index_map[ex.position]
def rebase_expr(expr, offset):
"""Subtract the given offset from each column access.
Assumes that named attribute references have been converted to integer
positions.
"""
assert offset > 0
for ex in expr.walk():
assert not isinstance(ex, NamedAttributeRef)
if isinstance(ex, UnnamedAttributeRef):
ex.position -= offset