encoding/csv: add option to reuse slices returned by Read

In many cases the records returned by Reader.Read will only be used between calls
to Read and become garbage once a new record is read. In this case, instead of
allocating a new slice on each call to Read, we can reuse the last allocated slice
for successive calls to avoid unnecessary allocations.

This change adds a new field ReuseRecord to the Reader struct to enable this reuse.

ReuseRecord is false by default to avoid breaking existing code which dependss on
the current behaviour.

I also added 4 new benchmarks, corresponding to the existing Read benchmarks, which
set ReuseRecord to true.

Benchstat on my local machine (old is ReuseRecord = false, new is ReuseRecord = true)

name                          old time/op    new time/op    delta
Read-8                          2.75µs ± 2%    1.88µs ± 1%  -31.52%  (p=0.000 n=14+15)
ReadWithFieldsPerRecord-8       2.75µs ± 0%    1.89µs ± 1%  -31.43%  (p=0.000 n=13+13)
ReadWithoutFieldsPerRecord-8    2.77µs ± 1%    1.88µs ± 1%  -32.06%  (p=0.000 n=15+15)
ReadLargeFields-8               55.4µs ± 1%    54.2µs ± 0%   -2.07%  (p=0.000 n=15+14)

name                          old alloc/op   new alloc/op   delta
Read-8                            664B ± 0%       24B ± 0%  -96.39%  (p=0.000 n=15+15)
ReadWithFieldsPerRecord-8         664B ± 0%       24B ± 0%  -96.39%  (p=0.000 n=15+15)
ReadWithoutFieldsPerRecord-8      664B ± 0%       24B ± 0%  -96.39%  (p=0.000 n=15+15)
ReadLargeFields-8               3.94kB ± 0%    2.98kB ± 0%  -24.39%  (p=0.000 n=15+15)

name                          old allocs/op  new allocs/op  delta
Read-8                            18.0 ± 0%       8.0 ± 0%  -55.56%  (p=0.000 n=15+15)
ReadWithFieldsPerRecord-8         18.0 ± 0%       8.0 ± 0%  -55.56%  (p=0.000 n=15+15)
ReadWithoutFieldsPerRecord-8      18.0 ± 0%       8.0 ± 0%  -55.56%  (p=0.000 n=15+15)
ReadLargeFields-8                 24.0 ± 0%      12.0 ± 0%  -50.00%  (p=0.000 n=15+15)

Fixes #19721

Change-Id: I79b14128bb9bb3465f53f40f93b1b528a9da6f58
Reviewed-on: https://go-review.googlesource.com/41730
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

1 files changed, 31 insertions, 0 deletions

diff --git a/src/encoding/csv/reader_test.go b/src/encoding/csv/reader_test.go
index 7b3aca4c5f..5ab1b61256 100644
--- a/src/encoding/csv/reader_test.go
+++ b/src/encoding/csv/reader_test.go
@@ -24,6 +24,7 @@ var readTests = []struct {
 	LazyQuotes       bool
 	TrailingComma    bool
 	TrimLeadingSpace bool
+	ReuseRecord      bool
 
 	Error  string
 	Line   int // Expected error line if != 0
@@ -260,6 +261,15 @@ x,,,
 			{"c", "d", "e"},
 		},
 	},
+	{
+		Name:        "ReadAllReuseRecord",
+		ReuseRecord: true,
+		Input:       "a,b\nc,d",
+		Output: [][]string{
+			{"a", "b"},
+			{"c", "d"},
+		},
+	},
 }
 
 func TestRead(t *testing.T) {
@@ -274,6 +284,7 @@ func TestRead(t *testing.T) {
 		r.LazyQuotes = tt.LazyQuotes
 		r.TrailingComma = tt.TrailingComma
 		r.TrimLeadingSpace = tt.TrimLeadingSpace
+		r.ReuseRecord = tt.ReuseRecord
 		if tt.Comma != 0 {
 			r.Comma = tt.Comma
 		}
@@ -369,3 +380,23 @@ xxxxxxxxxxxxxxxxxxxxxxxx,yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy,zzzzzzzzzzzzzzzzzzzzzz
 xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx,yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy,zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz,wwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww,vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
 `, 3))
 }
+
+func BenchmarkReadReuseRecord(b *testing.B) {
+	benchmarkRead(b, func(r *Reader) { r.ReuseRecord = true }, benchmarkCSVData)
+}
+
+func BenchmarkReadReuseRecordWithFieldsPerRecord(b *testing.B) {
+	benchmarkRead(b, func(r *Reader) { r.ReuseRecord = true; r.FieldsPerRecord = 4 }, benchmarkCSVData)
+}
+
+func BenchmarkReadReuseRecordWithoutFieldsPerRecord(b *testing.B) {
+	benchmarkRead(b, func(r *Reader) { r.ReuseRecord = true; r.FieldsPerRecord = -1 }, benchmarkCSVData)
+}
+
+func BenchmarkReadReuseRecordLargeFields(b *testing.B) {
+	benchmarkRead(b, func(r *Reader) { r.ReuseRecord = true }, strings.Repeat(`xxxxxxxxxxxxxxxx,yyyyyyyyyyyyyyyy,zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz,wwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww,vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
+xxxxxxxxxxxxxxxxxxxxxxxx,yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy,zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz,wwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww,vvvv
+,,zzzz,wwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww,vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
+xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx,yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy,zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz,wwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwwww,vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
+`, 3))
+}