From d733b63d6d662e76fca6ea377268cf3a8182c073 Mon Sep 17 00:00:00 2001
From: Vsevolod Tolstopyatov <qwwdfsad@gmail.com>
Date: Wed, 21 Aug 2024 13:55:58 +0200
Subject: [PATCH 1/2] Minor cleanup and simplification of readString

* Request Long.MAX_VALUE from the source at once (note: no need for a loop as we are unable to build the string of such size anyway)
* Get rid of redundant checks in commonReadUtf8

Ticks one of the boxes in #342
---
 benchmarks/README.md                          | 11 ++++-
 .../commonMain/kotlin/ReadStringBenchmark.kt  | 42 +++++++++++++++++++
 core/common/src/Utf8.kt                       | 17 +++-----
 3 files changed, 57 insertions(+), 13 deletions(-)
 create mode 100644 benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt

diff --git a/benchmarks/README.md b/benchmarks/README.md
index 49ba39a0d..dc2917d35 100644
--- a/benchmarks/README.md
+++ b/benchmarks/README.md
@@ -8,4 +8,13 @@ Currently, the suite includes benchmarks on:
 - segment pooling performance.
 
 The suite doesn't include benchmarks for more complex APIs inherited from Okio as these APIs are subject to change.
-Such benchmarks will be added later along with corresponding changes in the library.
\ No newline at end of file
+Such benchmarks will be added later along with corresponding changes in the library.
+
+### Quickstart
+
+For JVM:
+```
+./gradlew :kotlinx-io-benchmarks:jvmJar
+
+java -jar benchmarks/build/benchmarks/jvm/jars/kotlinx-io-benchmarks-jvm-jmh-0.6.0-SNAPSHOT-JMH.jar  ReadStringBenchmark -f 1 -wi 5 -i 5 -tu us -w 1 -r 1
+```
\ No newline at end of file
diff --git a/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt b/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt
new file mode 100644
index 000000000..694f6bf1f
--- /dev/null
+++ b/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt
@@ -0,0 +1,42 @@
+package kotlinx.io.benchmarks
+
+import kotlinx.benchmark.Benchmark
+import kotlinx.benchmark.Param
+import kotlinx.benchmark.Scope
+import kotlinx.benchmark.Setup
+import kotlinx.benchmark.State
+import kotlinx.io.Buffer
+import kotlinx.io.Source
+import kotlinx.io.readCodePointValue
+import kotlinx.io.readString
+import kotlinx.io.writeCodePointValue
+import kotlinx.io.writeString
+import kotlin.random.Random
+
+
+@State(Scope.Benchmark)
+open class ReadStringBenchmark() {
+
+    @Param("16", "64", "512")
+    var size: Int = 0
+
+    val buffer: Buffer = Buffer()
+
+    @Setup
+    fun setup() {
+        val string = buildString { repeat(size) { append(('a'..'z').random()) } }
+        buffer.writeString(string)
+    }
+
+
+    @Benchmark
+    fun bufferReadString(): String {
+        return buffer.copy().readString()
+    }
+
+    @Benchmark
+    fun sourceReadString(): String {
+        val source: Source = buffer.copy()
+        return source.readString()
+    }
+}
diff --git a/core/common/src/Utf8.kt b/core/common/src/Utf8.kt
index d4e7202d5..f455fbce3 100644
--- a/core/common/src/Utf8.kt
+++ b/core/common/src/Utf8.kt
@@ -206,10 +206,7 @@ public fun Sink.writeString(chars: CharSequence, startIndex: Int = 0, endIndex:
  */
 @OptIn(InternalIoApi::class)
 public fun Source.readString(): String {
-    var req: Long = Segment.SIZE.toLong()
-    while (request(req)) {
-        req *= 2
-    }
+    request(Long.MAX_VALUE) // Request all data
     return buffer.commonReadUtf8(buffer.size)
 }
 
@@ -607,10 +604,7 @@ private fun Buffer.commonWriteUtf8CodePoint(codePoint: Int) {
 
 @OptIn(UnsafeIoApi::class)
 private fun Buffer.commonReadUtf8(byteCount: Long): String {
-    require(byteCount >= 0 && byteCount <= Int.MAX_VALUE) {
-        "byteCount ($byteCount) is not within the range [0..${Int.MAX_VALUE})"
-    }
-    require(byteCount)
+    // Invariant: byteCount was request()'ed into this buffer beforehand
     if (byteCount == 0L) return ""
 
     UnsafeBufferOperations.iterate(this) { ctx, head ->
@@ -624,7 +618,6 @@ private fun Buffer.commonReadUtf8(byteCount: Long): String {
             }
         }
     }
-
-    // If the string spans multiple segments, delegate to readBytes().
-    return readByteArray(byteCount.toInt()).commonToUtf8String()
-}
+        // If the string spans multiple segments, delegate to readBytes()
+        return readByteArray(byteCount.toInt()).commonToUtf8String()
+    }

From 49e16cc408c6564d7b5100a4ff34140fe9a0803a Mon Sep 17 00:00:00 2001
From: Vsevolod Tolstopyatov <qwwdfsad@gmail.com>
Date: Fri, 23 Aug 2024 16:30:24 +0200
Subject: [PATCH 2/2] ~comment

---
 benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt b/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt
index 694f6bf1f..1b1d15b98 100644
--- a/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt
+++ b/benchmarks/src/commonMain/kotlin/ReadStringBenchmark.kt
@@ -17,7 +17,7 @@ import kotlin.random.Random
 @State(Scope.Benchmark)
 open class ReadStringBenchmark() {
 
-    @Param("16", "64", "512")
+    @Param("16", "64", "512") // Fits into a single segment, so the benchmark does not measure segment boundaries crossing
     var size: Int = 0
 
     val buffer: Buffer = Buffer()