Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Parquet metadata udfs #33

Closed
wants to merge 1 commit into from
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions src/lib.rs
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,7 @@ use pgrx::{prelude::*, GucContext, GucFlags, GucRegistry};

mod arrow_parquet;
mod parquet_copy_hook;
mod parquet_udfs;
mod pgrx_utils;
mod type_compat;

Expand Down
2 changes: 2 additions & 0 deletions src/parquet_udfs.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
pub(crate) mod metadata;
pub(crate) mod schema;
268 changes: 268 additions & 0 deletions src/parquet_udfs/metadata.rs
Original file line number Diff line number Diff line change
@@ -0,0 +1,268 @@
use ::parquet::file::statistics::Statistics;
use pgrx::{iter::TableIterator, name, pg_extern, pg_schema};

use crate::arrow_parquet::uri_utils::{parquet_metadata_from_uri, parse_uri, uri_as_string};

#[pg_schema]
mod parquet {
use super::*;

#[pg_extern]
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

high-level pgrx question, how does it handle function signature changes over time?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

cargo pgrx schema generates sql file of the current version. Then, we need to use a schema diff tool to generate the schema diffs. An example script that does it example new release.

#[allow(clippy::type_complexity)]
fn metadata(
uri: String,
) -> TableIterator<
'static,
(
name!(uri, String),
name!(row_group_id, i64),
name!(row_group_num_rows, i64),
name!(row_group_num_columns, i64),
name!(row_group_bytes, i64),
name!(column_id, i64),
name!(file_offset, i64),
name!(num_values, i64),
name!(path_in_schema, String),
name!(type_name, String),
name!(stats_null_count, Option<i64>),
name!(stats_distinct_count, Option<i64>),
name!(stats_min, Option<String>),
name!(stats_max, Option<String>),
name!(compression, String),
name!(encodings, String),
name!(index_page_offset, Option<i64>),
name!(dictionary_page_offset, Option<i64>),
name!(data_page_offset, i64),
name!(total_compressed_size, i64),
name!(total_uncompressed_size, i64),
),
> {
let uri = parse_uri(&uri);

// new_current_thread runtime uses the current thread to run the tokio reactor.
// This uses the same thread that is running the Postgres backend.
let runtime = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.unwrap_or_else(|e| {
panic!("Failed to create tokio runtime: {}", e);
});

let parquet_metadata = runtime.block_on(parquet_metadata_from_uri(&uri));

let mut rows = vec![];

for (row_group_id, row_group) in parquet_metadata.row_groups().iter().enumerate() {
let row_group_num_rows = row_group.num_rows();
let row_group_num_columns = row_group.num_columns() as i64;
let row_group_bytes = row_group.total_byte_size();

for (column_id, column) in row_group.columns().iter().enumerate() {
let file_offset = column.file_offset();

let num_values = column.num_values();

let path_in_schema = column.column_path().string();

let type_name = column.column_type().to_string();

let mut stats_min = None;
let mut stats_max = None;
let mut stats_null_count = None;
let mut stats_distinct_count = None;

if let Some(statistics) = column.statistics() {
stats_min = stats_min_value_to_str(statistics);

stats_max = stats_max_value_to_str(statistics);

stats_null_count = statistics.null_count_opt().map(|v| v as i64);

stats_distinct_count = statistics.distinct_count_opt().map(|v| v as i64);
}

let compression = column.compression().to_string();

let encodings = column
.encodings()
.iter()
.map(|e| e.to_string())
.collect::<Vec<_>>()
.join(",");

let index_page_offset = column.index_page_offset();

let dictionary_page_offset = column.dictionary_page_offset();

let data_page_offset = column.data_page_offset();

let total_compressed_size = column.compressed_size();

let total_uncompressed_size = column.uncompressed_size();

let row = (
uri_as_string(&uri),
row_group_id as i64,
row_group_num_rows,
row_group_num_columns,
row_group_bytes,
column_id as i64,
file_offset,
num_values,
path_in_schema,
type_name,
stats_null_count,
stats_distinct_count,
stats_min,
stats_max,
compression,
encodings,
index_page_offset,
dictionary_page_offset,
data_page_offset,
total_compressed_size,
total_uncompressed_size,
);

rows.push(row);
}
}

TableIterator::new(rows)
}

#[pg_extern]
fn file_metadata(
uri: String,
) -> TableIterator<
'static,
(
name!(uri, String),
name!(created_by, Option<String>),
name!(num_rows, i64),
name!(num_row_groups, i64),
name!(format_version, String),
),
> {
let uri = parse_uri(&uri);

// new_current_thread runtime uses the current thread to run the tokio reactor.
// This uses the same thread that is running the Postgres backend.
let runtime = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.unwrap_or_else(|e| {
panic!("Failed to create tokio runtime: {}", e);
});

let parquet_metadata = runtime.block_on(parquet_metadata_from_uri(&uri));

let created_by = parquet_metadata
.file_metadata()
.created_by()
.map(|c| c.to_string());

let num_rows = parquet_metadata.file_metadata().num_rows();

let num_row_groups = parquet_metadata.num_row_groups() as i64;

let format_version = parquet_metadata.file_metadata().version().to_string();

let row = (
uri_as_string(&uri),
created_by,
num_rows,
num_row_groups,
format_version,
);

TableIterator::new(vec![row])
}

#[pg_extern]
fn kv_metadata(
uri: String,
) -> TableIterator<
'static,
(
name!(uri, String),
name!(key, Vec<u8>),
name!(value, Option<Vec<u8>>),
),
> {
let uri = parse_uri(&uri);

// new_current_thread runtime uses the current thread to run the tokio reactor.
// This uses the same thread that is running the Postgres backend.
let runtime = tokio::runtime::Builder::new_current_thread()
.enable_all()
.build()
.unwrap_or_else(|e| {
panic!("Failed to create tokio runtime: {}", e);
});

let parquet_metadata = runtime.block_on(parquet_metadata_from_uri(&uri));
let kv_metadata = parquet_metadata.file_metadata().key_value_metadata();

if kv_metadata.is_none() {
return TableIterator::new(vec![]);
}

let kv_metadata = kv_metadata.expect("kv_metadata should be Some");

let mut rows = vec![];

for kv in kv_metadata {
let key = kv.key.as_bytes().to_owned();
let value = kv.value.as_ref().map(|v| v.as_bytes().to_owned());

let row = (uri_as_string(&uri), key, value);

rows.push(row);
}

TableIterator::new(rows)
}
}

fn stats_min_value_to_str(statistics: &Statistics) -> Option<String> {
match &statistics {
Statistics::Boolean(val_stats) => val_stats.min_opt().map(|v| v.to_string()),
Statistics::Int32(val_stats) => val_stats.min_opt().map(|v| v.to_string()),
Statistics::Int64(val_stats) => val_stats.min_opt().map(|v| v.to_string()),
Statistics::Int96(val_stats) => val_stats.min_opt().map(|v| v.to_string()),
Statistics::Float(val_stats) => val_stats.min_opt().map(|v| v.to_string()),
Statistics::Double(val_stats) => val_stats.min_opt().map(|v| v.to_string()),
Statistics::ByteArray(val_stats) => val_stats.min_opt().map(|v| match v.as_utf8() {
Ok(v) => v.to_string(),
Err(_) => v.to_string(),
}),
Statistics::FixedLenByteArray(val_stats) => {
val_stats.min_opt().map(|v| match v.as_utf8() {
Ok(v) => v.to_string(),
Err(_) => v.to_string(),
})
}
}
}

fn stats_max_value_to_str(statistics: &Statistics) -> Option<String> {
match statistics {
Statistics::Boolean(statistics) => statistics.max_opt().map(|v| v.to_string()),
Statistics::Int32(statistics) => statistics.max_opt().map(|v| v.to_string()),
Statistics::Int64(statistics) => statistics.max_opt().map(|v| v.to_string()),
Statistics::Int96(statistics) => statistics.max_opt().map(|v| v.to_string()),
Statistics::Float(statistics) => statistics.max_opt().map(|v| v.to_string()),
Statistics::Double(statistics) => statistics.max_opt().map(|v| v.to_string()),
Statistics::ByteArray(statistics) => statistics.max_opt().map(|v| match v.as_utf8() {
Ok(v) => v.to_string(),
Err(_) => v.to_string(),
}),
Statistics::FixedLenByteArray(statistics) => {
statistics.max_opt().map(|v| match v.as_utf8() {
Ok(v) => v.to_string(),
Err(_) => v.to_string(),
})
}
}
}
Loading
Loading