harmos_storage/query/transform/
project.rs1pub use harmos_domain::model::storage::query::transform::project::{Project, Projection};
17
18use crate::storage::query::transform::FieldPathExecution;
21use std::sync::Arc;
23use arrow_array::RecordBatch;
25use arrow_schema::Schema;
26use super::{FieldPath, invalid};
28use crate::Result;
29
30pub trait ProjectExecution {
43 fn output_schema(&self, input: &Schema) -> Result<Schema>;
44
45 fn apply(&self, input: &RecordBatch, check: &mut impl FnMut() -> Result<()>) -> Result<RecordBatch>;
46}
47
48impl ProjectExecution for Project {
51 fn output_schema(&self, input: &Schema) -> Result<Schema> {
52 if self.fields.is_empty() || self.fields.len() > 4096 {
53 return Err(invalid("projection requires bounded nonempty fields"));
54 }
55 let mut fields = Vec::with_capacity(self.fields.len());
56 for projected in &self.fields {
57 if projected.name.is_empty() || self.fields.iter().filter(|field| field.name == projected.name).count() != 1 {
58 return Err(invalid("projection names must be nonempty and unique"));
59 }
60 let field = projected.field.field(input)?.with_name(&projected.name);
61 fields.push(field);
62 }
63 Ok(Schema::new_with_metadata(fields, input.metadata().clone()))
64 }
65
66 fn apply(&self, input: &RecordBatch, check: &mut impl FnMut() -> Result<()>) -> Result<RecordBatch> {
67 let schema = Arc::new(self.output_schema(input.schema().as_ref())?);
68 let mut columns = Vec::with_capacity(self.fields.len());
69 for projected in &self.fields {
70 check()?;
71 let mut array = Arc::clone(projected.field.array(input)?);
73 for depth in 1..projected.field.fields.len() {
74 let parent = FieldPath { fields: projected.field.fields[..depth].to_vec() };
75 let parent = parent.array(input)?;
76 if let Some(nulls) = parent.nulls() {
77 let combined = arrow_buffer::NullBuffer::union(array.nulls(), Some(nulls));
78 let data = array
79 .to_data()
80 .into_builder()
81 .nulls(combined)
82 .build()
83 .map_err(|_| invalid("projected parent validity is invalid"))?;
84 array = arrow_array::make_array(data);
85 }
86 }
87 columns.push(array);
88 }
89 RecordBatch::try_new(schema, columns).map_err(|_| invalid("projected batch is invalid"))
90 }
91}
92
93