Merge pull request #35 from prrao87/zstd-parquet

Zstd parquet

data/create_edges_follows.py

@@ -97,7 +97,7 @@ def main() -> None:
         edges_df = edges_df.head(NUM)
         print(f"Limiting edges to {NUM} per the `--num` argument")
     # Write nodes
-    edges_df.write_parquet(Path("output/edges") / "follows.parquet", compression="snappy")
+    edges_df.write_parquet(Path("output/edges") / "follows.parquet")
     print(f"Wrote {len(edges_df)} edges for {len(persons_df)} persons")
 
 

data/create_edges_interests.py

@@ -51,7 +51,7 @@ def main() -> None:
         print(f"Limiting edges to {NUM} per the `--num` argument")
     # Write nodes
     edges_df = edges_df.rename({"id": "from", "interests": "to"})
-    edges_df.write_parquet(Path("output/edges") / "interests.parquet", compression="snappy")
+    edges_df.write_parquet(Path("output/edges") / "interests.parquet")
     print(f"Wrote {len(edges_df)} edges for {len(persons_df)} persons")
 
 

data/create_edges_location.py

@@ -51,7 +51,6 @@ def main() -> None:
     # Write nodes
     edges_df = edges_df.rename({"city_id": "to", "id": "from"}).write_parquet(
         Path("output/edges") / "lives_in.parquet",
-        compression="snappy",
     )
     print(f"Generated residence cities for persons. Top 5 common cities are: {', '.join(top_5)}")
 

data/create_edges_location_city_state.py

@@ -26,7 +26,7 @@ def main() -> None:
         .rename({"city_id": "from", "state_id": "to"})
     )
     # Write nodes
-    edges_df.write_parquet(Path("output/edges") / "city_in.parquet", compression="snappy")
+    edges_df.write_parquet(Path("output/edges") / "city_in.parquet")
     print(f"Wrote {len(edges_df)} edges for {len(cities_df)} cities")
 
 

data/create_edges_location_state_country.py

@@ -22,7 +22,7 @@ def main() -> None:
         .rename({"state_id": "from", "country_id": "to"})
     )
     # Write nodes
-    edges_df.write_parquet(Path("output/edges") / "state_in.parquet", compression="snappy")
+    edges_df.write_parquet(Path("output/edges") / "state_in.parquet")
     print(f"Wrote {len(edges_df)} edges for {len(states_df)} states")
 
 

data/create_nodes_interests.py

@@ -20,7 +20,6 @@ def main(filename: str) -> pl.DataFrame:
     # Write to csv
     interests_df.select(pl.col("id"), pl.all().exclude("id")).write_parquet(
         Path("output/nodes") / "interests.parquet",
-        compression="snappy",
     )
     print(f"Wrote {interests_df.shape[0]} interests nodes to parquet")
     return interests

data/create_nodes_location.py

@@ -59,7 +59,6 @@ def write_city_nodes(cities_of_interest: pl.DataFrame) -> pl.DataFrame:
     # Write to csv
     city_nodes.select(pl.col("id"), pl.all().exclude("id")).write_parquet(
         Path("output/nodes") / "cities.parquet",
-        compression="snappy",
     )
     print(f"Wrote {city_nodes.shape[0]} cities to parquet")
     return city_nodes

data/create_nodes_person.py

@@ -62,7 +62,6 @@ def main() -> None:
     # Write nodes
     persons_df.select(pl.col("id"), pl.all().exclude("id")).write_parquet(
         Path("output/nodes") / "persons.parquet",
-        compression="snappy",
     )
     print(f"Wrote {persons_df.shape[0]} person nodes to parquet")
 

kuzudb/benchmark_query.py

@@ -1,5 +1,5 @@
 """
-Use the `pytest-benchmark` library to more formally benchmark the Neo4j queries with warmup and iterations.
+Use the `pytest-benchmark` library to more formally benchmark the Kùzu queries with warmup and iterations.
 `pip install pytest-benchmark`
 """
 import pytest

kuzudb/query.py

@@ -1,5 +1,5 @@
 """
-Run a series of queries on the Neo4j database
+Run a series of queries on an existing Kùzu database
 """
 from typing import Any
 

requirements.txt

@@ -2,7 +2,7 @@ faker~=19.12.0
 polars~=0.19.0
 numpy>=1.25.0
 pyarrow~=13.0.0
-kuzu==0.0.11
+kuzu==0.0.12
 neo4j~=5.13.0
 python-dotenv>=1.0.0
 codetiming>=1.4.0