Refactored concatenating

experiment.py

@@ -37,7 +37,6 @@
 )
 from klinker.blockers.hybrid import (
     CompositeRelationalDeepBlocker,
-    CompositeLightEABlocker,
     CompositeEmbeddingBlocker,
 )
 from klinker.blockers.base import Blocker
@@ -559,12 +558,7 @@ def process_pipeline(
     tracker = experiment_info.tracker
 
     start = time.time()
-    blocks = blocker.assign(
-        left=klinker_dataset.left,
-        right=klinker_dataset.right,
-        left_rel=klinker_dataset.left_rel,
-        right_rel=klinker_dataset.right_rel,
-    )
+    blocks = blocker.assign_from_dataset(klinker_dataset)
     blocks.to_parquet(experiment_info.blocks_path, overwrite=True)
 
     end = time.time()
@@ -1373,65 +1367,6 @@ def composite_relational_deepblocker(
     return (blocker, click.get_current_context().params, end - start)
 
 
-@cli.command
-@click.option("--min-token-length", type=int, default=3)
-@click.option("--top-n-a", type=int, default=None)
-@click.option("--top-n-r", type=int, default=None)
-@click.option("--depth", type=int, default=2)
-@click.option("--mini-dim", type=int, default=16)
-@click.option("--rel-dim", type=int)
-@embedding_options
-def composite_light_ea_blocker(
-    min_token_length: int,
-    top_n_a: Optional[int],
-    top_n_r: Optional[int],
-    depth: int,
-    mini_dim: int,
-    rel_dim: Optional[int],
-    inner_encoder: Type[TokenizedFrameEncoder],
-    embeddings: str,
-    inner_encoder_batch_size: int,
-    reduce_dim_to: int,
-    reduce_sample_perc: float,
-    block_builder: Type[EmbeddingBlockBuilder],
-    block_builder_kwargs: str,
-    n_neighbors: int,
-    n_candidates: Optional[int],
-    force: bool,
-    save_emb: bool,
-):
-    if top_n_a and top_n_a < 0:
-        top_n_a = None
-    if top_n_r and top_n_r < 0:
-        top_n_r = None
-    inner_encoder_inst = create_inner_encoder(
-        inner_encoder,
-        embeddings,
-        inner_encoder_batch_size,
-        reduce_dim_to,
-        reduce_sample_perc,
-    )
-    bb_kwargs = parse_bb_kwargs(
-        block_builder_kwargs, n_neighbors, block_builder, n_candidates
-    )
-    print(bb_kwargs)
-    start = time.time()
-    blocker = CompositeLightEABlocker(
-        min_token_length=min_token_length,
-        top_n_a=top_n_a,
-        top_n_r=top_n_r,
-        depth=depth,
-        mini_dim=mini_dim,
-        inner_encoder=inner_encoder_inst,
-        embedding_block_builder=block_builder,
-        embedding_block_builder_kwargs=bb_kwargs,
-        force=force,
-        save=save_emb,
-    )
-    end = time.time()
-    return (blocker, click.get_current_context().params, end - start)
-
-
 @cli.command()
 @click.option("--min-token-length", type=int, default=3)
 @click.option("--top-n-a", type=int, default=None)
@@ -1680,9 +1615,4 @@ def composite_relational_token_clustering_lsh_blocker(
 
 
 if __name__ == "__main__":
-    # import dask
-    # dask.config.set(scheduler='processes')
-    # from dask.distributed import Client, LocalCluster
-    # cluster = LocalCluster()  # Launches a scheduler and workers locally
-    # client = Client(cluster)  # Connect to distributed cluster and override default
     cli()

src/klinker/__init__.py

@@ -2,18 +2,12 @@
 
 from .data import (
     KlinkerBlockManager,
-    KlinkerDaskFrame,
     KlinkerDataset,
-    KlinkerFrame,
     NNBasedKlinkerBlockManager,
     CompositeWithNNBasedKlinkerBlockManager,
 )
 
 __all__ = [
-    "KlinkerFrame",
-    "KlinkerPandasFrame",
-    "KlinkerDaskFrame",
-    "KlinkerPandasTripleFrame",
     "KlinkerDataset",
     "KlinkerBlockManager",
     "NNBasedKlinkerBlockManager",