From 902519a26bd0546157f0d03afbe77eaff477a658 Mon Sep 17 00:00:00 2001
From: Pavel Punegov <32519625+PavelPunegov@users.noreply.github.com>
Date: Wed, 10 Jan 2018 14:57:04 +0300
Subject: [PATCH] Check all whitespace characters according to Unicode standard

---
 backend.native/tests/build.gradle         |  5 +++
 backend.native/tests/runtime/text/trim.kt | 49 +++++++++++++++++++++++
 runtime/src/main/cpp/KString.cpp          | 25 +++---------
 3 files changed, 60 insertions(+), 19 deletions(-)
 create mode 100644 backend.native/tests/runtime/text/trim.kt

diff --git a/backend.native/tests/build.gradle b/backend.native/tests/build.gradle
index c87f0721f54..7ffd3dd2ca4 100644
--- a/backend.native/tests/build.gradle
+++ b/backend.native/tests/build.gradle
@@ -1512,6 +1512,11 @@ task to_string0(type: RunKonanTest) {
     source = "runtime/text/to_string0.kt"
 }
 
+task trim(type: RunKonanTest) {
+    goldValue = "OK\n"
+    source = "runtime/text/trim.kt"
+}
+
 task chars0(type: RunKonanTest) {
     expectedFail = (project.testTarget == 'wasm32') // Uses exceptions.
     source = "runtime/text/chars0.kt"
diff --git a/backend.native/tests/runtime/text/trim.kt b/backend.native/tests/runtime/text/trim.kt
new file mode 100644
index 00000000000..be6251d3ffd
--- /dev/null
+++ b/backend.native/tests/runtime/text/trim.kt
@@ -0,0 +1,49 @@
+package runtime.text.trim
+
+import kotlin.test.*
+
+/**
+ * Tests correct conversions of floats/doubles along with trimming of leading/trailing whitespaces.
+ * String.trim() trims all whitespaces (see kotlin/text/Strings.kt) while String.toFloat() uses
+ * the same approach as java.lang.Float.parseFloat()
+ */
+@Test fun runTest() {
+    convertToFloatingPoint()
+    convertWithWhitespaces()
+    trimWhitespaces()
+
+    println("OK")
+}
+
+private fun convertToFloatingPoint() {
+    assertEquals(expected = 3.14F, actual = " 3.14  ".toFloat(), message = "String float should be trimmed")
+    assertEquals(expected = 3.14, actual = " 3.14  ".toDouble(), message = "String double should be trimmed")
+    assertEquals(expected = 7.15F, actual = "\u0019 7.15  ".toFloat(), message = "String float should be trimmed")
+    assertEquals(expected = 42.3, actual = "\n 42.3 ".toDouble(), message = "String double should be trimmed")
+}
+
+private fun convertWithWhitespaces() {
+    val s = "\u0009 \u000A 2.71 \u000D"
+    assertEquals(expected = 2.71F, actual = s.toFloat(),
+            message = "String should be cleared of LF, CR, TAB and converted to Float")
+    assertEquals(expected = 2.71, actual = s.toDouble(),
+            message = "String should be cleared of LF, CR, TAB and converted to Double")
+
+    // Special symbols should not be trimmed during String to Float/Double conversion
+    assertFailsWith<NumberFormatException> { "\u202F3.14".toFloat() }
+    assertFailsWith<NumberFormatException> { "\u20293.14".toDouble() }
+    assertFailsWith<NumberFormatException> { "3.14\u200B".toDouble() }
+    assertFailsWith<NumberFormatException> { "3.14\u200B ABC".toDouble() }
+}
+
+private fun trimWhitespaces() {
+    assertEquals(expected = "String", actual = "  String".trim(), message = "Trim leading spaces")
+    assertEquals(expected = "String  ", actual = "    String  ".trimStart(), message = "Trim start")
+    assertEquals(expected = "  String", actual = "  String \t ".trimEnd(), message = "Trim end")
+
+    assertEquals(expected = "String", actual = "\u0020 \u202FString\u2028\u2029".trim(),
+            message = "Trim special whitespaces")
+    assertEquals(expected = "\u1FFFString", actual = "\u0085  \u1FFFString".trim(),
+            message = "Trim special whitespace but should left a unicode symbol")
+    assertEquals(expected = "String\tSTR", actual = " \nString\tSTR  ".trim(), message = "Trim newline")
+}
diff --git a/runtime/src/main/cpp/KString.cpp b/runtime/src/main/cpp/KString.cpp
index 5dcd9be4920..e00e690798a 100644
--- a/runtime/src/main/cpp/KString.cpp
+++ b/runtime/src/main/cpp/KString.cpp
@@ -671,17 +671,17 @@ int iswalnum_Konan(KChar ch) {
 }
 
 int iswspace_Konan(KChar ch) {
-  // Optimized case for ASCII.
+  // FILE, GROUP, RECORD, UNIT separators, # Zs SPACE or # Cc CONTROLs
   if ((ch >= 0x1c && ch <= 0x20) || (ch >= 0x9 && ch <= 0xd)) {
     return true;
   }
-  if (ch == 0x1680) {
-    return true;
-  }
-  if (ch < 0x2000 || ch == 0x2007) {
+  // not (# Zs OGHAM SPACE MARK or # Cc or # Zs NO-BREAK SPACE)
+  if (ch < 0x2000 && !(ch == 0x1680 || ch == 0x85 || ch == 0xA0)) {
     return false;
   }
-  return ch <= 0x200b || ch == 0x2028 || ch == 0x2029 || ch == 0x3000;
+  // if # Zl LINE SEPARATOR or # Zp PARAGRAPH SEPARATOR or # Zs NARROW NO-BREAK SPACE
+  // or # Zs MEDIUM MATHEMATICAL SPACE or # Zs IDEOGRAPHIC SPACE
+  return ch < 0x200b || ch == 0x2028 || ch == 0x2029 || ch == 0x202F || ch == 0x205F || ch == 0x3000;
 }
 
 int iswupper_Konan(KChar ch) {
@@ -708,19 +708,6 @@ int iswlower_Konan(KChar ch) {
   return getType(ch) == LOWERCASE_LETTER;
 }
 
-void checkParsingErrors(const char* c_str, const char* end, KStdString::size_type c_str_size) {
-  if (end == c_str) {
-    ThrowNumberFormatException();
-  }
-  // According to http://docs.oracle.com/javase/8/docs/api/java/lang/Double.html#valueOf-java.lang.String-
-  // trailing whitespace characters must be ignored so we need to do an additional check.
-  for (const char* p = end; p < c_str + c_str_size; p++) {
-    if (!iswspace_Konan(*p)) {
-      ThrowNumberFormatException();
-    }
-  }
-}
-
 } // namespace
 
 extern "C" {