sarahfujimori
commented
1 year ago - Add global variable for supported aggregation types and for table prefixes. The table prefixes dictionary is needed because DuckDB plans currently don't include the table name in the parquet scan nodes. In the TPC-H schema, we can tell which table a column belongs to by looking at the column name (e.g. 'l_orderkey' has the prefix 'l' and is in the table 'lineitem'), so for now we use this dictionary to determine the table name of the parquet scan.
- An earlier version of DuckDB labels parquet scan nodes as PARQUET_SCAN and newer versions label it READ_PARQUET - update code to support both of these
- Bug fix for parsing filter nodes