Struct StreamingSort 

pub struct StreamingSort { /* private fields */ }

External merge-sort for large datasets that don’t fit in memory.

Phase 1: collect buffer_size records at a time, sort each run, write to a temporary file. Phase 2: k-way merge of all temporary run files.

Example

use scirs2_io::streaming::relational::StreamingSort;
use serde_json::json;

let mut sorter = StreamingSort::new(
    16,                                              // buffer size
    |v| v["name"].as_str().unwrap_or("").to_string(), // sort key
    true,                                            // ascending
);

let records = vec![
    json!({"name": "Charlie"}),
    json!({"name": "Alice"}),
    json!({"name": "Bob"}),
];

let sorted = sorter.sort_all(records).expect("sort");
assert_eq!(sorted[0]["name"], "Alice");

Implementations

impl StreamingSort

pub fn new<F>(buffer_size: usize, key_fn: F, ascending: bool) -> Self

where F: Fn(&Value) -> String + Send + 'static,

Create a streaming sorter.

• buffer_size: number of records to buffer per run before writing to disk.
• key_fn: extract the sort key from each record.
• ascending: sort direction.

pub fn sort_all(&mut self, records: Vec<Value>) -> Result<Vec<Value>>

Sort all records using external merge-sort.

For datasets small enough to fit in buffer_size, this is equivalent to an in-memory sort.