From 16cf4ca2cae528d5bdc310f011017f1063c4ce7b Mon Sep 17 00:00:00 2001
From: Fabian Schmidt <fabschmidt96@gmail.com>
Date: Tue, 23 Jul 2024 16:30:50 +0200
Subject: [PATCH] moved from reading String to reading bytes. A little faster,
 still need to implement for multithreaded solution

---
 src/main/rust/src/bin/single_thread.rs | 20 +++++++--------
 src/main/rust/src/lib.rs               | 35 ++++++++++++++++++++++++++
 2 files changed, 45 insertions(+), 10 deletions(-)

diff --git a/src/main/rust/src/bin/single_thread.rs b/src/main/rust/src/bin/single_thread.rs
index 6037fec..26add47 100644
--- a/src/main/rust/src/bin/single_thread.rs
+++ b/src/main/rust/src/bin/single_thread.rs
@@ -1,10 +1,9 @@
-use std::{
-    fs::File,
-    io::{BufRead, BufReader},
-};
 use std::collections::HashMap;
+use std::fs::File;
+use std::io::{BufReader, Read};
 use std::time::Instant;
 
+use onebrc::{parse_line, read_bytes_until};
 
 const DEFAULT_HASHMAP_LENGTH: usize = 10000;
 
@@ -14,12 +13,13 @@ fn main() {
         HashMap::with_capacity(DEFAULT_HASHMAP_LENGTH);
 
     let file = File::open("../../../measurements.txt").expect("File measurements.txt not found");
-    let reader = BufReader::new(file);
-    for line_result in reader.lines() {
+    let mut bytes = BufReader::new(file).bytes();
+    while let Some(line_result) = read_bytes_until(&mut bytes, b'\n') {
         let line = line_result.expect("could not read line");
-        let (station, temp) = line.split_once(';').unwrap();
-        let temp = onebrc::parse_temp(temp.as_bytes());
-        let measurements_option = stations.get_mut(station);
+        let (station, temp) = parse_line(&line);
+        let station = unsafe { String::from_utf8_unchecked(Vec::from(station)) };
+        let temp = onebrc::parse_temp(temp);
+        let measurements_option = stations.get_mut(&station);
         if let Some(measurements) = measurements_option {
             measurements.update(temp);
         } else {
@@ -29,7 +29,7 @@ fn main() {
                 count: 1,
                 sum: temp,
             };
-            stations.insert(station.to_owned(), measurements);
+            stations.insert(station, measurements);
         }
     }
     let mut stations: Vec<String> = stations.iter().map(|(station, measurements)| {
diff --git a/src/main/rust/src/lib.rs b/src/main/rust/src/lib.rs
index 694dbd5..9d44b10 100644
--- a/src/main/rust/src/lib.rs
+++ b/src/main/rust/src/lib.rs
@@ -1,4 +1,6 @@
 use std::fmt::Display;
+use std::fs::File;
+use std::io::{BufReader, Bytes};
 
 #[derive(Copy, Clone)]
 pub struct StationMeasurements {
@@ -64,4 +66,37 @@ pub fn parse_temp(bytes: &[u8]) -> isize {
     } else {
         as_decimal as isize
     }
+}
+
+#[inline]
+pub fn read_bytes_until(bytes: &mut Bytes<BufReader<File>>, delimiter: u8) -> Option<std::io::Result<[u8; 108]>> {
+    // 108 max length of line in bytes
+    let mut buf: [u8; 108] = [b'#'; 108];
+    let mut idx = 0;
+    while let Some(byte) = bytes.next() {
+        if byte.is_err() {
+            panic!("Could not read byte");
+        }
+        let byte = byte.unwrap();
+        if delimiter == byte {
+            return Some(Ok(buf));
+        }
+        buf[idx] = byte;
+        idx += 1;
+    }
+    None
+}
+
+#[inline]
+pub fn parse_line(line: &[u8]) -> (&[u8], &[u8]) {
+    let mut idx = 0;
+    while idx < line.len() && line[idx] != b';' {
+        idx += 1;
+    }
+    let station = &line[0..idx];
+    let midpoint = idx + 1;
+    while idx < line.len() && line[idx] != b'#' {
+        idx += 1;
+    }
+    (station, &line[midpoint..idx])
 }
\ No newline at end of file