Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
pola-rs
GitHub Repository: pola-rs/polars
Path: blob/main/crates/polars-parquet/src/parquet/statistics/fixed_len_binary.rs
6940 views
1
use polars_parquet_format::Statistics as ParquetStatistics;
2
3
use crate::parquet::error::{ParquetError, ParquetResult};
4
use crate::parquet::schema::types::PrimitiveType;
5
6
#[derive(Debug, Clone, PartialEq)]
7
pub struct FixedLenStatistics {
8
pub primitive_type: PrimitiveType,
9
pub null_count: Option<i64>,
10
pub distinct_count: Option<i64>,
11
pub max_value: Option<Vec<u8>>,
12
pub min_value: Option<Vec<u8>>,
13
}
14
15
impl FixedLenStatistics {
16
pub fn deserialize(
17
v: &ParquetStatistics,
18
size: usize,
19
primitive_type: PrimitiveType,
20
) -> ParquetResult<Self> {
21
if let Some(ref v) = v.max_value {
22
if v.len() != size {
23
return Err(ParquetError::oos(
24
"The max_value of statistics MUST be plain encoded",
25
));
26
}
27
};
28
if let Some(ref v) = v.min_value {
29
if v.len() != size {
30
return Err(ParquetError::oos(
31
"The min_value of statistics MUST be plain encoded",
32
));
33
}
34
};
35
36
Ok(Self {
37
primitive_type,
38
null_count: v.null_count,
39
distinct_count: v.distinct_count,
40
max_value: v.max_value.clone().map(|mut x| {
41
x.truncate(size);
42
x
43
}),
44
min_value: v.min_value.clone().map(|mut x| {
45
x.truncate(size);
46
x
47
}),
48
})
49
}
50
51
pub fn serialize(&self) -> ParquetStatistics {
52
ParquetStatistics {
53
null_count: self.null_count,
54
distinct_count: self.distinct_count,
55
max_value: self.max_value.clone(),
56
min_value: self.min_value.clone(),
57
max: None,
58
min: None,
59
is_max_value_exact: None,
60
is_min_value_exact: None,
61
}
62
}
63
}
64
65