Merge pull request #8 from PSeitz/mainbrunch

display bench throughput, fix build
datalust · Sep 18, 2023 · 8183504 · 8183504
2 parents c03bd4d + 9124a2f
commit 8183504
Show file tree

Hide file tree

Showing 5 changed files with 34 additions and 9 deletions.
diff --git a/benches/mod.rs b/benches/mod.rs
@@ -10,6 +10,7 @@ use std::str;
 fn read_10kb_event_stacktrace_offsets_simd(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| Document::scan_trusted(input))
 }
 
@@ -18,6 +19,7 @@ fn read_10kb_event_stacktrace_offsets_simd_from_const_parts(b: &mut test::Benche
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
     let const_offsets = Document::scan_trusted(input).into_offsets();
 
+    b.bytes = input.len() as u64;
     b.iter(|| unsafe { const_offsets.to_document_unchecked(input) })
 }
 
@@ -26,6 +28,7 @@ fn read_10kb_event_stacktrace_offsets_simd_reuse(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
     let mut detached = Some(Document::scan_trusted(input).detach());
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let reuse = detached.take().unwrap();
 
@@ -40,13 +43,15 @@ fn read_10kb_event_stacktrace_offsets_simd_reuse(b: &mut test::Bencher) {
 fn read_10kb_event_stacktrace_offsets_fallback(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| Document::scan_trusted_fallback(input))
 }
 
 #[bench]
 fn read_10kb_event_stacktrace_value_serde_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let v: serde_json::Value = serde_json::from_slice(input).unwrap();
         v
@@ -57,6 +62,7 @@ fn read_10kb_event_stacktrace_value_serde_json(b: &mut test::Bencher) {
 fn read_10kb_event_stacktrace_value_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let v: json::JsonValue = json::parse(str::from_utf8(input).unwrap()).unwrap();
         v
@@ -67,6 +73,7 @@ fn read_10kb_event_stacktrace_value_json(b: &mut test::Bencher) {
 fn read_10kb_event_stacktrace_value_simd_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let mut input = input.to_vec();
         let v = simd_json::to_borrowed_value(&mut input).unwrap();
@@ -78,20 +85,23 @@ fn read_10kb_event_stacktrace_value_simd_json(b: &mut test::Bencher) {
 fn read_10kb_event_stacktrace_value_to_vec(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| input.to_vec())
 }
 
 #[bench]
 fn read_10kb_event_stacktrace_split(b: &mut test::Bencher) {
     let input = include_str!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| input.split('"').for_each(drop))
 }
 
 #[bench]
 fn read_10kb_event_stacktrace_validate_utf8(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| str::from_utf8(input).unwrap())
 }
 
@@ -100,6 +110,7 @@ fn read_10kb_event_stacktrace_validate_utf8(b: &mut test::Bencher) {
 fn read_10kb_event_stacktrace_offsets_simd_to_serde_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| Document::scan_trusted(input).to_value())
 }
 
@@ -110,13 +121,15 @@ fn convert_10kb_event_stacktrace_offsets_to_serde_json(b: &mut test::Bencher) {
 
     let doc = Document::scan_trusted(input);
 
+    b.bytes = input.len() as u64;
     b.iter(|| doc.to_value())
 }
 
 #[bench]
 fn read_10kb_event_stacktrace_offsets_simd_sparse(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let doc = Document::scan_trusted(input);
 
@@ -137,6 +150,7 @@ fn read_10kb_event_stacktrace_offsets_simd_sparse(b: &mut test::Bencher) {
 fn read_10kb_event_stacktrace_serde_json_sparse(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/10kb_event_stacktrace.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let v: serde_json::Value = serde_json::from_slice(input).unwrap();
 
@@ -169,6 +183,7 @@ fn unescape_10kb_event_stacktrace(b: &mut test::Bencher) {
         .as_str()
         .unwrap();
 
+    b.bytes = input.len() as u64;
     b.iter(|| stacktrace.to_unescaped())
 }
 
@@ -187,6 +202,7 @@ fn unescape_10kb_event_stacktrace_to_string(b: &mut test::Bencher) {
         .as_str()
         .unwrap();
 
+    b.bytes = input.len() as u64;
     b.iter(|| stacktrace.as_raw().to_owned())
 }
 
@@ -195,27 +211,31 @@ fn iter_top_level_entries_600b_event_no_escape(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
     let document = Document::scan_trusted(input);
 
+    b.bytes = input.len() as u64;
     b.iter(|| document.as_map().entries().for_each(drop))
 }
 
 #[bench]
 fn read_600b_event_no_escape_offsets_simd(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| Document::scan_trusted(input))
 }
 
 #[bench]
 fn read_600b_event_no_escape_offsets_fallback(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| Document::scan_trusted_fallback(input))
 }
 
 #[bench]
 fn read_600b_event_no_escape_value_serde_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let v: serde_json::Value = serde_json::from_slice(input).unwrap();
         v
@@ -226,6 +246,7 @@ fn read_600b_event_no_escape_value_serde_json(b: &mut test::Bencher) {
 fn read_600b_event_no_escape_value_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let v: json::JsonValue = json::parse(str::from_utf8(input).unwrap()).unwrap();
         v
@@ -236,6 +257,7 @@ fn read_600b_event_no_escape_value_json(b: &mut test::Bencher) {
 fn read_600b_event_no_escape_value_simd_json(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| {
         let mut input = input.to_vec();
         let v = simd_json::to_borrowed_value(&mut input).unwrap();
@@ -247,19 +269,22 @@ fn read_600b_event_no_escape_value_simd_json(b: &mut test::Bencher) {
 fn read_600b_event_no_escape_value_to_vec(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| input.to_vec())
 }
 
 #[bench]
 fn read_600b_event_no_escape_split(b: &mut test::Bencher) {
     let input = include_str!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| input.split('"').for_each(drop))
 }
 
 #[bench]
 fn read_600b_event_no_escape_validate_utf8(b: &mut test::Bencher) {
     let input = include_bytes!("../cases/600b_event_no_escape.json");
 
+    b.bytes = input.len() as u64;
     b.iter(|| str::from_utf8(input).unwrap())
 }
diff --git a/src/lib.rs b/src/lib.rs
@@ -37,8 +37,8 @@ mod macros;
 
 mod std_ext;
 
-mod unescape;
 pub mod de;
+mod unescape;
 pub use de::Document;
 
 #[cfg(test)]

diff --git a/src/std_ext.rs b/src/std_ext.rs
@@ -4,5 +4,5 @@ This module has some "standard" extensions used in the deserializer.
 These extensions follow the same layout as Rust's standard library.
 */
 
-pub(crate) mod char;
 pub(crate) mod arch;
+pub(crate) mod char;
diff --git a/src/std_ext/arch/aarch64.rs b/src/std_ext/arch/aarch64.rs
@@ -1,7 +1,5 @@
 #[cfg(any(target_feature = "neon", target_feature = "crc"))]
-use std::{
-    arch::aarch64::*,
-};
+use std::arch::aarch64::*;
 
 #[target_feature(enable = "neon")]
 #[inline]
@@ -31,6 +29,6 @@ pub unsafe fn vmovemask_u8(a: uint8x8_t) -> u8 {
             0b0010_0000,
             0b0100_0000,
             0b1000_0000,
-        ])
+        ]),
     ))
 }
diff --git a/src/unescape.rs b/src/unescape.rs
@@ -191,7 +191,7 @@ impl<'a> ScanFnInput<'a> {
 
 #[inline(always)]
 fn interest_unescape<'a, I: BorrowMut<ScanFnInput<'a>>>(mut i: I) {
-    let mut i = i.borrow_mut();
+    let i = i.borrow_mut();
 
     let escaped = i.scan.escape;
     i.scan.escape = !escaped;
@@ -244,8 +244,10 @@ fn interest_unescape<'a, I: BorrowMut<ScanFnInput<'a>>>(mut i: I) {
                         match i.scan.first_surrogate.take() {
                             // if we had a surrogate pair, then attempt to map it to a multibyte
                             Some(first) => {
-                                let ch = crate::std_ext::char::try_from_utf16_surrogate_pair(first, code)
-                                    .map_err(|_| ())?;
+                                let ch = crate::std_ext::char::try_from_utf16_surrogate_pair(
+                                    first, code,
+                                )
+                                .map_err(|_| ())?;
                                 i.push_unescaped_char(ch);
                             }
                             // if we didn't have a surrogate pair,