Module filter 

Vectorized predicate evaluation on dictionary-encoded columns.

Implements late materialization for DictEncoded columns: predicates are evaluated on compact integer IDs rather than decompressed strings. This turns O(N * string_len) evaluation into O(dict_size * string_len + N) — a large win when cardinality is low (the common case for dict-encoded columns).

Short-circuit paths

• eq, not-in-dict: return all-zero mask immediately (O(1) reject)
• ne, not-in-dict: return all-ones mask immediately (O(1) accept)
• contains, empty-match-set: return all-zero mask immediately

Bitmask encoding

Results are returned as Vec<u64> packed bitmasks: bit i is set if row i passes the predicate. Use words_for(row_count) to size the output.

Functions

bitmask_all

All-ones bitmask for row_count rows.

bitmask_and

Bitwise AND of two equal-length bitmasks.

decoded_dict_eval_contains

Evaluate a substring-contains predicate on a DecodedColumn::DictEncoded.

decoded_dict_eval_eq

Evaluate an equality predicate on a DecodedColumn::DictEncoded.

decoded_dict_eval_like

Evaluate a LIKE predicate on a DecodedColumn::DictEncoded.

decoded_dict_eval_ne

Evaluate a not-equal predicate on a DecodedColumn::DictEncoded.

dict_eval_contains

Evaluate a substring-contains predicate on a memtable DictEncoded column.

dict_eval_eq

Evaluate an equality predicate on a memtable DictEncoded column.

dict_eval_like

Evaluate a LIKE predicate on a memtable DictEncoded column.

dict_eval_ne

Evaluate a not-equal predicate on a memtable DictEncoded column.

words_for

Number of u64 words needed to hold row_count bits.