Common isWellFormed() implementation (#3829)

wavesplatform · Jul 17, 2023 · b773589 · b773589
1 parent 961e643
commit b773589
Show file tree

Hide file tree

Showing 7 changed files with 52 additions and 36 deletions.
diff --git a/lang/js/src/main/scala/com/wavesplatform/lang/Global.scala b/lang/js/src/main/scala/com/wavesplatform/lang/Global.scala
@@ -120,9 +120,8 @@ object Global extends BaseGlobal {
       if (value.toLong.toDouble == value && value - 1 < Long.MaxValue) BD.valueOf(value.toLong)
       else BD.valueOf(value)
 
-    Right(BigInt(decimal
-      .setScale(scale.toInt, round.mode)
-      .unscaledValue))
+    val scaled = decimal.setScale(scale.toInt, round.mode).unscaledValue
+    Right(BigInt(scaled))
   }
 
   implicit class BigIntOps(val v: BigInteger) extends AnyVal {
@@ -143,7 +142,4 @@ object Global extends BaseGlobal {
 
   override def ecrecover(messageHash: Array[Byte], signature: Array[Byte]): Array[Byte] =
     ???
-
-  override def isIllFormed(s: String): Boolean =
-    false
 }
diff --git a/lang/jvm/src/main/scala/com/wavesplatform/lang/Global.scala b/lang/jvm/src/main/scala/com/wavesplatform/lang/Global.scala
@@ -1,11 +1,7 @@
 package com.wavesplatform.lang
 
-import java.math.{BigInteger, MathContext, BigDecimal => BD}
-import java.security.spec.InvalidKeySpecException
-
-import cats.syntax.either._
+import cats.syntax.either.*
 import ch.obermuhlner.math.big.BigDecimalMath
-import com.google.common.base.Utf8
 import com.google.common.io.BaseEncoding
 import com.wavesplatform.common.merkle.Merkle
 import com.wavesplatform.common.utils.{Base58, Base64}
@@ -14,11 +10,13 @@ import com.wavesplatform.lang.v1.BaseGlobal
 import com.wavesplatform.lang.v1.evaluator.ctx.impl.Rounding
 import com.wavesplatform.lang.v1.evaluator.ctx.impl.crypto.RSA
 import com.wavesplatform.lang.v1.evaluator.ctx.impl.crypto.RSA.DigestAlgorithm
-import com.wavesplatform.zwaves.bls12.{Groth16 => Bls12Groth16}
-import com.wavesplatform.zwaves.bn256.{Groth16 => Bn256Groth16}
+import com.wavesplatform.zwaves.bls12.Groth16 as Bls12Groth16
+import com.wavesplatform.zwaves.bn256.Groth16 as Bn256Groth16
 import org.web3j.crypto.Sign
 import org.web3j.crypto.Sign.SignatureData
 
+import java.math.{BigInteger, MathContext, BigDecimal as BD}
+import java.security.spec.InvalidKeySpecException
 import scala.annotation.tailrec
 import scala.util.Try
 
@@ -82,8 +80,8 @@ object Global extends BaseGlobal {
   private val longContext    = new MathContext(longDigits)
   private val oldLongContext = MathContext.DECIMAL128
 
-  private val bigIntDigits = 154
-  private val bigMathContext = new MathContext(bigIntDigits)
+  private val bigIntDigits      = 154
+  private val bigMathContext    = new MathContext(bigIntDigits)
   private val oldBigMathContext = new MathContext(156 + 40)
 
   // Math functions
@@ -125,10 +123,8 @@ object Global extends BaseGlobal {
 
   def powBigInt(b: BigInt, bp: Long, e: BigInt, ep: Long, rp: Long, round: Rounding, useNewPrecision: Boolean): Either[String, BigInt] =
     tryEither {
-      val base    = toJBig(b, bp)
-      val exp     = toJBig(e, ep)
-
-
+      val base = toJBig(b, bp)
+      val exp  = toJBig(e, ep)
 
       val context = if (useNewPrecision) bigMathContext else oldBigMathContext
       val res = if (exp == BigDecimal(0.5).bigDecimal) {
@@ -187,7 +183,4 @@ object Global extends BaseGlobal {
     val pk = Sign.signedMessageHashToKey(messageHash, signatureData)
     base16Encoder.decode(pk.toString(16))
   }
-
-  override def isIllFormed(s: String): Boolean =
-    Try(Utf8.encodedLength(s)).isFailure
 }
diff --git a/lang/shared/src/main/scala/com/wavesplatform/lang/package.scala b/lang/shared/src/main/scala/com/wavesplatform/lang/package.scala
@@ -1,13 +1,13 @@
 package com.wavesplatform
 
-import cats.{Eval, Id, Monad, StackSafeMonad}
 import cats.data.EitherT
+import cats.{Eval, Id, Monad, StackSafeMonad}
 import monix.eval.Coeval
 
 package object lang {
   implicit def toError(msg: String): CommonError = CommonError(msg)
 
-  type ExecutionLog   = String
+  type ExecutionLog = String
 
   type CoevalF[F[_], A]               = Coeval[F[A]]
   type EvalF[F[_], A]                 = Eval[F[A]]
@@ -20,4 +20,20 @@ package object lang {
     override def pure[A](x: A): Coeval[A] =
       Coeval.now(x)
   }
+
+  implicit class StringOps(val s: String) extends AnyVal {
+    def isWellFormed: Boolean = {
+      var i = 0
+      var wellFormed = true
+      while (i < s.length && wellFormed) {
+        val c = s.charAt(i)
+        if (Character.isSurrogate(c)) {
+          if (s.codePointAt(i) == c) wellFormed = false
+          else i += 1
+        }
+        i += 1
+      }
+      wellFormed
+    }
+  }
 }
diff --git a/lang/shared/src/main/scala/com/wavesplatform/lang/v1/BaseGlobal.scala b/lang/shared/src/main/scala/com/wavesplatform/lang/v1/BaseGlobal.scala
@@ -385,8 +385,6 @@ trait BaseGlobal {
       }
     }
   }
-
-  def isIllFormed(s: String): Boolean
 }
 
 object BaseGlobal {

diff --git a/lang/shared/src/main/scala/com/wavesplatform/lang/v1/compiler/ExpressionCompiler.scala b/lang/shared/src/main/scala/com/wavesplatform/lang/v1/compiler/ExpressionCompiler.scala
@@ -3,7 +3,6 @@ package com.wavesplatform.lang.v1.compiler
 import cats.implicits.*
 import cats.{Id, Show}
 import com.wavesplatform.common.state.ByteStr
-import com.wavesplatform.lang.CommonError
 import com.wavesplatform.lang.v1.compiler.CompilationError.*
 import com.wavesplatform.lang.v1.compiler.CompilerContext.*
 import com.wavesplatform.lang.v1.compiler.Terms.*
@@ -28,14 +27,13 @@ import com.wavesplatform.lang.v1.parser.Expressions.{
 import com.wavesplatform.lang.v1.parser.Parser.LibrariesOffset
 import com.wavesplatform.lang.v1.parser.{BinaryOperation, Expressions, Parser}
 import com.wavesplatform.lang.v1.task.imports.*
-import com.wavesplatform.lang.v1.{BaseGlobal, ContractLimits, FunctionHeader}
+import com.wavesplatform.lang.v1.{ContractLimits, FunctionHeader}
+import com.wavesplatform.lang.{CommonError, StringOps}
 
 import java.nio.charset.StandardCharsets
 import scala.util.Try
 
 object ExpressionCompiler {
-  private val global: BaseGlobal = com.wavesplatform.lang.Global
-
   case class CompilationStepResultExpr(
       ctx: CompilerContext,
       expr: Terms.EXPR,
@@ -126,7 +124,7 @@ object ExpressionCompiler {
 
       def adjustStr(expr: Expressions.CONST_STRING, str: String): Either[CompilationError, CompilationStepResultExpr] =
         CONST_STRING(str)
-          .filterOrElse(_ => allowIllFormedStrings || !global.isIllFormed(str), CommonError(s"String '$str' contains ill-formed characters"))
+          .filterOrElse(_ => allowIllFormedStrings || str.isWellFormed, CommonError(s"String '$str' contains ill-formed characters"))
           .leftMap(e => CompilationError.Generic(expr.position.start, expr.position.end, e.message))
           .map(CompilationStepResultExpr(ctx, _, STRING, expr))
           .recover { case err => CompilationStepResultExpr(ctx, FAILED_EXPR(), NOTHING, expr, List(err)) }

diff --git a/lang/shared/src/main/scala/com/wavesplatform/lang/v1/evaluator/ctx/impl/PureContext.scala b/lang/shared/src/main/scala/com/wavesplatform/lang/v1/evaluator/ctx/impl/PureContext.scala
@@ -1023,7 +1023,7 @@ object PureContext {
       case CONST_STRING(m) :: CONST_STRING(sub) :: Nil =>
         Right {
           val i = m.indexOf(sub)
-          if (!global.isIllFormed(sub) && i != -1)
+          if (sub.isWellFormed && i != -1)
             CONST_LONG(m.codePointCount(0, i).toLong)
           else
             unit
@@ -1067,7 +1067,7 @@ object PureContext {
     ) {
       case CONST_STRING(m) :: CONST_STRING(sub) :: CONST_LONG(off) :: Nil =>
         val l = m.codePointCount(0, m.length)
-        Right(if (!global.isIllFormed(sub) && off >= 0 && off <= l) {
+        Right(if (sub.isWellFormed && off >= 0 && off <= l) {
           val i = m.indexOf(sub, m.offsetByCodePoints(0, off.toInt))
           if (i != -1) {
             CONST_LONG(m.codePointCount(0, i).toLong)
@@ -1113,7 +1113,7 @@ object PureContext {
       case CONST_STRING(m) :: CONST_STRING(sub) :: Nil =>
         Right({
           val i = m.lastIndexOf(sub)
-          if (!global.isIllFormed(sub) && i != -1) {
+          if (sub.isWellFormed && i != -1) {
             CONST_LONG(m.codePointCount(0, i).toLong)
           } else {
             unit
@@ -1161,7 +1161,7 @@ object PureContext {
         Right(if (off >= 0) {
           val offset = Math.min(off, m.codePointCount(0, m.length)).toInt
           val i      = m.lastIndexOf(sub, m.offsetByCodePoints(0, offset))
-          if (!global.isIllFormed(sub) && i != -1) {
+          if (sub.isWellFormed && i != -1) {
             CONST_LONG(m.codePointCount(0, i).toLong)
           } else {
             unit
@@ -1205,7 +1205,7 @@ object PureContext {
 
   private def split(str: String, sep: String, unicode: Boolean): Iterable[CONST_STRING] = {
     if (str == "") listWithEmptyStr
-    else if (unicode && global.isIllFormed(sep)) List(CONST_STRING(str).explicitGet())
+    else if (unicode && !sep.isWellFormed) List(CONST_STRING(str).explicitGet())
     else if (sep == "")
       if (unicode) {
         (1 to str.codePointCount(0, str.length))

diff --git a/lang/tests-js/src/test/scala/com/wavesplatform/JsAPITest.scala b/lang/tests-js/src/test/scala/com/wavesplatform/JsAPITest.scala
@@ -222,5 +222,20 @@ object JsAPITest extends JsTestBase {
       JSON.stringify(result.exprAst.expr) ==> JSON.stringify(JSON.parse(expected))
       JSON.stringify(result.errorList) ==> "[]"
     }
+
+    test("ill-formed characters") {
+      val script =
+        """
+          |{-# STDLIB_VERSION 6 #-}
+          |{-# CONTENT_TYPE DAPP #-}
+          |{-# SCRIPT_TYPE ACCOUNT #-}
+          |
+          |func call(a: String, b: Int) = {
+          |    let zzz = "aaa\ud87ebbb"
+          |    ([], zzz)
+          |}
+        """.stripMargin
+      assertCompileError(script, "contains ill-formed characters")
+    }
   }
 }
-Original file line number
+Diff line change
@@ Expand Up / @@ -385,8 +385,6 @@ trait BaseGlobal { @@
           }
         }
       }
-      def isIllFormed(s: String): Boolean
     }
     object BaseGlobal {
@@ Expand Down @@