From f9a467828a3925f2ca50bf58843a04f67c92c252 Mon Sep 17 00:00:00 2001
From: tyb0807 <sontuan.vu119@gmail.com>
Date: Mon, 30 Dec 2024 20:22:27 +0100
Subject: [PATCH] IntSeqEncoder trait and PlainEncoder implementation (#239)

This trait is for implementing different encoding algorithms to compress posting lists
(or any sorted integer sequences)
---
 Cargo.lock                        |  1 +
 rs/compression/Cargo.toml         |  1 +
 rs/compression/src/compression.rs | 20 +++++++++++++
 rs/compression/src/lib.rs         |  2 ++
 rs/compression/src/noc/mod.rs     |  1 +
 rs/compression/src/noc/noc.rs     | 48 +++++++++++++++++++++++++++++++
 6 files changed, 73 insertions(+)
 create mode 100644 rs/compression/src/compression.rs
 create mode 100644 rs/compression/src/noc/mod.rs
 create mode 100644 rs/compression/src/noc/noc.rs

diff --git a/Cargo.lock b/Cargo.lock
index d185c41..c0ee294 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -477,6 +477,7 @@ dependencies = [
  "env_logger",
  "log",
  "tempdir",
+ "utils",
 ]
 
 [[package]]
diff --git a/rs/compression/Cargo.toml b/rs/compression/Cargo.toml
index b125eff..bef3812 100644
--- a/rs/compression/Cargo.toml
+++ b/rs/compression/Cargo.toml
@@ -9,3 +9,4 @@ bitvec = "1"
 env_logger.workspace = true
 log.workspace = true
 tempdir.workspace = true
+utils.workspace = true
diff --git a/rs/compression/src/compression.rs b/rs/compression/src/compression.rs
new file mode 100644
index 0000000..56425fb
--- /dev/null
+++ b/rs/compression/src/compression.rs
@@ -0,0 +1,20 @@
+use std::fs::File;
+use std::io::BufWriter;
+
+use anyhow::Result;
+
+pub trait IntSeqEncoder {
+    /// Creates an encoder
+    fn new_encoder(universe: Option<usize>, size: usize) -> Self
+    where
+        Self: Sized;
+
+    /// Compresses a sorted slice of integers
+    fn encode(&mut self, values: &[u64]) -> Result<()>;
+
+    /// Returns the number of elements in the sequence
+    fn len(&self) -> usize;
+
+    /// Writes to disk and return number of bytes written.
+    fn write(&self, writer: &mut BufWriter<&mut File>) -> Result<usize>;
+}
diff --git a/rs/compression/src/lib.rs b/rs/compression/src/lib.rs
index e75354e..ee5e21e 100644
--- a/rs/compression/src/lib.rs
+++ b/rs/compression/src/lib.rs
@@ -1 +1,3 @@
+pub mod compression;
 pub mod elias_fano;
+pub mod noc;
diff --git a/rs/compression/src/noc/mod.rs b/rs/compression/src/noc/mod.rs
new file mode 100644
index 0000000..f7e789f
--- /dev/null
+++ b/rs/compression/src/noc/mod.rs
@@ -0,0 +1 @@
+pub mod noc;
diff --git a/rs/compression/src/noc/noc.rs b/rs/compression/src/noc/noc.rs
new file mode 100644
index 0000000..0c5cb5e
--- /dev/null
+++ b/rs/compression/src/noc/noc.rs
@@ -0,0 +1,48 @@
+use std::fs::File;
+use std::io::{BufWriter, Write};
+
+use anyhow::Result;
+use utils::io::wrap_write;
+
+use crate::compression::IntSeqEncoder;
+
+pub struct PlainEncoder {
+    size: usize,
+    sequence: Vec<u64>,
+}
+
+impl PlainEncoder {
+    pub fn new(size: usize) -> Self {
+        Self {
+            size,
+            sequence: Vec::new(),
+        }
+    }
+}
+
+impl IntSeqEncoder for PlainEncoder {
+    fn new_encoder(_universe: Option<usize>, size: usize) -> Self {
+        Self::new(size)
+    }
+
+    fn encode(&mut self, values: &[u64]) -> Result<()> {
+        self.sequence = values.to_vec();
+        Ok(())
+    }
+
+    fn len(&self) -> usize {
+        self.size
+    }
+
+    fn write(&self, writer: &mut BufWriter<&mut File>) -> Result<usize> {
+        let mut total_bytes_written = 0;
+
+        for &val in self.sequence.iter() {
+            total_bytes_written += wrap_write(writer, &val.to_le_bytes())?;
+        }
+
+        writer.flush()?;
+
+        Ok(total_bytes_written)
+    }
+}