utils: rewrite count_newlines using bytecount::count

This uses a optimized byte count and also makes use of SIMD instructions to optimize the processing of the byte arrays. Signed-off-by: Otavio Salvador <[email protected]>
jorickert · Oct 14, 2018 · e203057 · e203057
1 parent c4a8cdc
commit e203057
Show file tree

Hide file tree

Showing 4 changed files with 13 additions and 2 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -51,6 +51,7 @@ rustc-ap-rustc_target = "272.0.0"
 rustc-ap-syntax = "272.0.0"
 rustc-ap-syntax_pos = "272.0.0"
 failure = "0.1.1"
+bytecount = { version = "0.3", features = ["simd-accel"] }
 
 [dev-dependencies]
 assert_cli = "0.6"

diff --git a/src/lib.rs b/src/lib.rs
@@ -17,6 +17,7 @@
 #[macro_use]
 extern crate derive_new;
 extern crate atty;
+extern crate bytecount;
 extern crate diff;
 extern crate failure;
 extern crate itertools;

diff --git a/src/utils.rs b/src/utils.rs
@@ -10,6 +10,8 @@
 
 use std::borrow::Cow;
 
+use bytecount;
+
 use rustc_target::spec::abi;
 use syntax::ast::{
     self, Attribute, CrateSugar, MetaItem, MetaItemKind, NestedMetaItem, NestedMetaItemKind, Path,
@@ -305,8 +307,8 @@ pub fn stmt_expr(stmt: &ast::Stmt) -> Option<&ast::Expr> {
 
 #[inline]
 pub fn count_newlines(input: &str) -> usize {
-    // Using `as_bytes` to omit UTF-8 decoding
-    input.as_bytes().iter().filter(|&&c| c == b'\n').count()
+    // Using bytes to omit UTF-8 decoding
+    bytecount::count(input.as_bytes(), b'\n')
 }
 
 // For format_missing and last_pos, need to use the source callsite (if applicable).