def __init__(self, table_name, join_keys, df, join_name):
self.table = table_name
self.df = df
self.join_keys = [f"{table_name}.{k}" for k in join_keys]
self.df.columns = [f"{table_name}.{k}" for k in self.df.columns]
self.indicator_column = f"__in_{table_name}"
self.sample_columns = [
c for c in self.df.columns if c not in self.join_keys
] # exclude join keys
self.index_provider = rustlib.IndexProvider(
f"{join_name}/{table_name}.pk.indices", NULL)
log.info(f"DataTableActor of `{table_name}` is ready.")
def __init__(self, table, jct, join_spec):
self.jct = jct
self.table = table
parents = list(join_spec.join_tree.predecessors(table))
assert len(parents) <= 1, parents
if len(parents) == 1:
parent = parents[0]
join_keys = join_spec.join_graph[parent][table]["join_keys"]
self.table_join_key = f"{table}.{join_keys[table]}"
self.parent_join_key = f"{parent}.{join_keys[parent]}"
null_row_offset = self._insert_null_to_jct()
self.index_provider = rustlib.IndexProvider(
f"{join_spec.join_name}/{table}.jk.indices", null_row_offset)
else:
self.jct_distribution = get_distribution(
self.jct[f"{self.table}.weight"])
log.info(f"JoinCountTableActor `{table}` is ready.")