Path: blob/main/crates/polars-parquet/src/parquet/statistics/primitive.rs
6940 views
use polars_parquet_format::Statistics as ParquetStatistics;12use crate::parquet::error::{ParquetError, ParquetResult};3use crate::parquet::schema::types::PrimitiveType;4use crate::parquet::types;56#[derive(Debug, Clone, PartialEq)]7pub struct PrimitiveStatistics<T: types::NativeType> {8pub primitive_type: PrimitiveType,9pub null_count: Option<i64>,10pub distinct_count: Option<i64>,11pub min_value: Option<T>,12pub max_value: Option<T>,13}1415impl<T: types::NativeType> PrimitiveStatistics<T> {16pub fn deserialize(17v: &ParquetStatistics,18primitive_type: PrimitiveType,19) -> ParquetResult<Self> {20if v.max_value21.as_ref()22.is_some_and(|v| v.len() != size_of::<T>())23{24return Err(ParquetError::oos(25"The max_value of statistics MUST be plain encoded",26));27};28if v.min_value29.as_ref()30.is_some_and(|v| v.len() != size_of::<T>())31{32return Err(ParquetError::oos(33"The min_value of statistics MUST be plain encoded",34));35};3637Ok(Self {38primitive_type,39null_count: v.null_count,40distinct_count: v.distinct_count,41max_value: v.max_value.as_ref().map(|x| types::decode(x)),42min_value: v.min_value.as_ref().map(|x| types::decode(x)),43})44}4546pub fn serialize(&self) -> ParquetStatistics {47ParquetStatistics {48null_count: self.null_count,49distinct_count: self.distinct_count,50max_value: self.max_value.map(|x| x.to_le_bytes().as_ref().to_vec()),51min_value: self.min_value.map(|x| x.to_le_bytes().as_ref().to_vec()),52max: None,53min: None,54is_max_value_exact: None,55is_min_value_exact: None,56}57}58}596061