diff options
| author | Julien Cretel <jub0bsinthecloud@gmail.com> | 2025-05-13 11:11:13 +0000 |
|---|---|---|
| committer | Gopher Robot <gobot@golang.org> | 2025-05-14 11:07:32 -0700 |
| commit | 14fc54fc5735163abf7694b4efa2b7998c6696e2 (patch) | |
| tree | fa3fd0139de13691ed433f9e003d68146229c5d1 /src/strings | |
| parent | 9856afa77042d901288845f31c79885ba68da464 (diff) | |
| download | go-14fc54fc5735163abf7694b4efa2b7998c6696e2.tar.xz | |
bytes, strings: speed up Split{,After}Seq
CL 669735 brought a welcome performance boost to splitSeq; however, it rendered explodeSeq ineligible for inlining and failed to update that function's doc comment.
This CL inlines the call to explodeSeq in splitSeq, thereby unlocking a further speedup in the case of an empty separator, and removes function explodeSeq altogether.
Some benchmarks results:
goos: darwin
goarch: amd64
pkg: strings
cpu: Intel(R) Core(TM) i7-6700HQ CPU @ 2.60GHz
│ old │ new │
│ sec/op │ sec/op vs base │
SplitSeqEmptySeparator-8 5.136m ± 6% 3.180m ± 6% -38.09% (p=0.000 n=20)
SplitSeqSingleByteSeparator-8 995.9µ ± 1% 988.4µ ± 0% -0.75% (p=0.000 n=20)
SplitSeqMultiByteSeparator-8 593.1µ ± 2% 591.7µ ± 1% ~ (p=0.253 n=20)
SplitAfterSeqEmptySeparator-8 5.554m ± 3% 3.432m ± 2% -38.20% (p=0.000 n=20)
SplitAfterSeqSingleByteSeparator-8 997.4µ ± 0% 1000.0µ ± 8% ~ (p=0.121 n=20)
SplitAfterSeqMultiByteSeparator-8 591.7µ ± 1% 588.9µ ± 0% -0.48% (p=0.004 n=20)
geomean 1.466m 1.247m -14.97%
│ old │ new │
│ B/op │ B/op vs base │
SplitSeqEmptySeparator-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=20) ¹
SplitSeqSingleByteSeparator-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=20) ¹
SplitSeqMultiByteSeparator-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=20) ¹
SplitAfterSeqEmptySeparator-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=20) ¹
SplitAfterSeqSingleByteSeparator-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=20) ¹
SplitAfterSeqMultiByteSeparator-8 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=20) ¹
geomean ² +0.00% ²
¹ all samples are equal
² summaries must be >0 to compute geomean
Change-Id: I5767b68dc1a4fbcb2ac20683830a49ee3eb1bee1
GitHub-Last-Rev: 344934071f3220a1afea3def306dadfee720d311
GitHub-Pull-Request: golang/go#73685
Reviewed-on: https://go-review.googlesource.com/c/go/+/672175
Reviewed-by: qiu laidongfeng2 <2645477756@qq.com>
Reviewed-by: Robert Griesemer <gri@google.com>
Auto-Submit: Keith Randall <khr@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Keith Randall <khr@google.com>
Reviewed-by: Keith Randall <khr@golang.org>
Diffstat (limited to 'src/strings')
| -rw-r--r-- | src/strings/iter.go | 19 |
1 files changed, 7 insertions, 12 deletions
diff --git a/src/strings/iter.go b/src/strings/iter.go index a42e78ee09..69fe031739 100644 --- a/src/strings/iter.go +++ b/src/strings/iter.go @@ -31,23 +31,18 @@ func Lines(s string) iter.Seq[string] { } } -// explodeSeq returns an iterator over the runes in s. -func explodeSeq(s string, yield func(string) bool) { - for len(s) > 0 { - _, size := utf8.DecodeRuneInString(s) - if !yield(s[:size]) { - return - } - s = s[size:] - } -} - // splitSeq is SplitSeq or SplitAfterSeq, configured by how many // bytes of sep to include in the results (none or all). func splitSeq(s, sep string, sepSave int) iter.Seq[string] { return func(yield func(string) bool) { if len(sep) == 0 { - explodeSeq(s, yield) + for len(s) > 0 { + _, size := utf8.DecodeRuneInString(s) + if !yield(s[:size]) { + return + } + s = s[size:] + } return } for { |