mito2/sst/parquet/
stats.rs

1// Copyright 2023 Greptime Team
2//
3// Licensed under the Apache License, Version 2.0 (the "License");
4// you may not use this file except in compliance with the License.
5// You may obtain a copy of the License at
6//
7//     http://www.apache.org/licenses/LICENSE-2.0
8//
9// Unless required by applicable law or agreed to in writing, software
10// distributed under the License is distributed on an "AS IS" BASIS,
11// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12// See the License for the specific language governing permissions and
13// limitations under the License.
14
15//! Statistics of parquet SSTs.
16
17use std::borrow::Borrow;
18use std::collections::HashSet;
19use std::sync::Arc;
20
21use api::v1::SemanticType;
22use datafusion_common::pruning::PruningStatistics;
23use datafusion_common::{Column, ScalarValue};
24use datatypes::arrow::array::{ArrayRef, BooleanArray, UInt64Array};
25use parquet::file::metadata::RowGroupMetaData;
26use store_api::metadata::RegionMetadataRef;
27use store_api::storage::ColumnId;
28
29use crate::sst::parquet::format::{ReadFormat, StatValues};
30
31/// Statistics for pruning row groups.
32pub(crate) struct RowGroupPruningStats<'a, T> {
33    /// Metadata of SST row groups.
34    row_groups: &'a [T],
35    /// Helper to read the SST.
36    read_format: &'a ReadFormat,
37    /// The metadata of the region.
38    /// It contains the schema a query expects to read. If it is not None, we use it instead
39    /// of the metadata in the SST to get the column id of a column as the SST may have
40    /// different columns.
41    expected_metadata: Option<RegionMetadataRef>,
42    /// If true, skip columns with Field semantic type during pruning.
43    skip_fields: bool,
44}
45
46impl<'a, T> RowGroupPruningStats<'a, T> {
47    /// Creates a new statistics to prune specific `row_groups`.
48    pub(crate) fn new(
49        row_groups: &'a [T],
50        read_format: &'a ReadFormat,
51        expected_metadata: Option<RegionMetadataRef>,
52        skip_fields: bool,
53    ) -> Self {
54        Self {
55            row_groups,
56            read_format,
57            expected_metadata,
58            skip_fields,
59        }
60    }
61
62    /// Returns the column id of specific column name if we need to read it.
63    /// Prefers the column id in the expected metadata if it exists.
64    /// Returns None if skip_fields is true and the column is a Field.
65    fn column_id_to_prune(&self, name: &str) -> Option<ColumnId> {
66        let metadata = self
67            .expected_metadata
68            .as_ref()
69            .unwrap_or_else(|| self.read_format.metadata());
70        let col = metadata.column_by_name(name)?;
71
72        // Skip field columns when skip_fields is enabled
73        if self.skip_fields && col.semantic_type == SemanticType::Field {
74            return None;
75        }
76
77        Some(col.column_id)
78    }
79
80    /// Returns the default value of all row groups for `column` according to the metadata.
81    fn compat_default_value(&self, column: &str) -> Option<ArrayRef> {
82        let metadata = self.expected_metadata.as_ref()?;
83        let col_metadata = metadata.column_by_name(column)?;
84        col_metadata
85            .column_schema
86            .create_default_vector(self.row_groups.len())
87            .unwrap_or(None)
88            .map(|vector| vector.to_arrow_array())
89    }
90}
91
92impl<T: Borrow<RowGroupMetaData>> RowGroupPruningStats<'_, T> {
93    /// Returns the null count of all row groups for `column` according to the metadata.
94    fn compat_null_count(&self, column: &str) -> Option<ArrayRef> {
95        let metadata = self.expected_metadata.as_ref()?;
96        let col_metadata = metadata.column_by_name(column)?;
97        let value = col_metadata
98            .column_schema
99            .create_default()
100            .unwrap_or(None)?;
101        let values = self.row_groups.iter().map(|meta| {
102            if value.is_null() {
103                u64::try_from(meta.borrow().num_rows()).ok()
104            } else {
105                Some(0)
106            }
107        });
108        Some(Arc::new(UInt64Array::from_iter(values)))
109    }
110}
111
112impl<T: Borrow<RowGroupMetaData>> PruningStatistics for RowGroupPruningStats<'_, T> {
113    fn min_values(&self, column: &Column) -> Option<ArrayRef> {
114        let column_id = self.column_id_to_prune(&column.name)?;
115        match self.read_format.min_values(self.row_groups, column_id) {
116            StatValues::Values(values) => Some(values),
117            StatValues::NoColumn => self.compat_default_value(&column.name),
118            StatValues::NoStats => None,
119        }
120    }
121
122    fn max_values(&self, column: &Column) -> Option<ArrayRef> {
123        let column_id = self.column_id_to_prune(&column.name)?;
124        match self.read_format.max_values(self.row_groups, column_id) {
125            StatValues::Values(values) => Some(values),
126            StatValues::NoColumn => self.compat_default_value(&column.name),
127            StatValues::NoStats => None,
128        }
129    }
130
131    fn num_containers(&self) -> usize {
132        self.row_groups.len()
133    }
134
135    fn null_counts(&self, column: &Column) -> Option<ArrayRef> {
136        let column_id = self.column_id_to_prune(&column.name)?;
137        match self.read_format.null_counts(self.row_groups, column_id) {
138            StatValues::Values(values) => Some(values),
139            StatValues::NoColumn => self.compat_null_count(&column.name),
140            StatValues::NoStats => None,
141        }
142    }
143
144    fn row_counts(&self, _column: &Column) -> Option<ArrayRef> {
145        // TODO(LFC): Impl it.
146        None
147    }
148
149    fn contained(&self, _column: &Column, _values: &HashSet<ScalarValue>) -> Option<BooleanArray> {
150        // TODO(LFC): Impl it.
151        None
152    }
153}
mito2/sst/parquet/stats.rs

mito2/sst/parquet/
stats.rs
