Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/castorini/quackir/llms.txt

Use this file to discover all available pages before exploring further.

IndexType and SearchDB are the two configuration enums exported from the top-level quackir package. IndexType determines the schema and indexing strategy used when creating a table, while SearchDB identifies which database backend a workflow targets. 
from quackir import IndexType, SearchDB

​
IndexType

IndexType controls the column layout created by init_table and the insertion logic used by load_jsonl_table / load_parquet_table.
MemberValueDescription
IndexType.SPARSE'sparse'Creates an (id, contents) table and builds a BM25 full-text index.
IndexType.DENSE'dense'Creates an (id, embedding) table suitable for cosine similarity search.

​
Usage

from quackir import IndexType
from quackir.index import DuckDBIndexer

indexer = DuckDBIndexer("my.db")

# Sparse (BM25) table
indexer.init_table("sparse_corpus", IndexType.SPARSE)

# Dense (vector) table — default embedding dimension is 768
indexer.init_table("dense_corpus", IndexType.DENSE, embedding_dim=1024)
SQLite only supports IndexType.SPARSE. Passing IndexType.DENSE to SQLiteIndexer.init_table raises a ValueError.

​
Detection

All three indexers expose get_index_type(table_name) to auto-detect the type of an existing table by inspecting its column names:
  • A contents column → IndexType.SPARSE
  • An embedding column → IndexType.DENSE
detected = indexer.get_index_type("corpus")
print(detected)  # IndexType.SPARSE

​
SearchDB

SearchDB identifies the database engine. It is used by CLI tooling and environment variable loading helpers in quackir._base to select the appropriate indexer or searcher at runtime.
MemberValueDescription
SearchDB.DUCKDB'duckdb'DuckDB file-based database. Requires db_path.
SearchDB.SQLITE'sqlite'SQLite file-based database. Requires db_path.
SearchDB.POSTGRES'postgres'PostgreSQL server. Requires db_name and db_user.

​
Usage

from quackir import SearchDB

db_type = SearchDB.DUCKDB
print(db_type.value)  # 'duckdb'

# Branching on backend
if db_type == SearchDB.DUCKDB:
    from quackir.index import DuckDBIndexer
    indexer = DuckDBIndexer("duck.db")
elif db_type == SearchDB.SQLITE:
    from quackir.index import SQLiteIndexer
    indexer = SQLiteIndexer("sqlite.db")
elif db_type == SearchDB.POSTGRES:
    from quackir.index import PostgresIndexer
    indexer = PostgresIndexer(db_name="quackir", user="postgres")

Build docs developers (and LLMs) love

Get started for freeTalk to us