fix: std.parseJson reports truncated input cleanly

Author: He-Pin

bench/resources/go_suite/manifestTomlEx.jsonnet.golden

@@ -55,5 +55,5 @@
          }
       }
    },
-   "nothing": "\n\n[prometheusOperator]\n\n\n  [prometheusOperator.clusterRole]\n\n\n    [[prometheusOperator.clusterRole.rules]]\n      apiGroups = [\n        \"authentication.k8s.io\"\n      ]\n      resources = [\n        \"tokenreviews\"\n      ]\n      verbs = [\n        \"create\"\n      ]\n\n    [[prometheusOperator.clusterRole.rules]]\n      apiGroups = [\n        \"authorization.k8s.io\"\n      ]\n      resources = [\n        \"subjectaccessreviews\"\n      ]\n      verbs = [\n        \"create\"\n      ]\n\n  [prometheusOperator.service]\n\n\n    [prometheusOperator.service.spec]\n\n\n      [[prometheusOperator.service.spec.ports]]\n        name = \"https\"\n        port = 8443\n        targetPort = \"https\"\n\n  [prometheusOperator.serviceMonitor]\n\n\n    [prometheusOperator.serviceMonitor.spec]\n\n\n      [[prometheusOperator.serviceMonitor.spec.endpoints]]\n        bearerTokenFile = \"/var/run/secrets/kubernetes.io/serviceaccount/token\"\n        honorLabels = true\n        port = \"https\"\n        scheme = \"https\"\n\n        [prometheusOperator.serviceMonitor.spec.endpoints.tlsConfig]\n          insecureSkipVerify = true"
+   "nothing": "[prometheusOperator]\n\n  [prometheusOperator.clusterRole]\n\n    [[prometheusOperator.clusterRole.rules]]\n      apiGroups = [\n        \"authentication.k8s.io\"\n      ]\n      resources = [\n        \"tokenreviews\"\n      ]\n      verbs = [\n        \"create\"\n      ]\n\n    [[prometheusOperator.clusterRole.rules]]\n      apiGroups = [\n        \"authorization.k8s.io\"\n      ]\n      resources = [\n        \"subjectaccessreviews\"\n      ]\n      verbs = [\n        \"create\"\n      ]\n\n  [prometheusOperator.service]\n\n    [prometheusOperator.service.spec]\n\n      [[prometheusOperator.service.spec.ports]]\n        name = \"https\"\n        port = 8443\n        targetPort = \"https\"\n\n  [prometheusOperator.serviceMonitor]\n\n    [prometheusOperator.serviceMonitor.spec]\n\n      [[prometheusOperator.serviceMonitor.spec.endpoints]]\n        bearerTokenFile = \"/var/run/secrets/kubernetes.io/serviceaccount/token\"\n        honorLabels = true\n        port = \"https\"\n        scheme = \"https\"\n\n        [prometheusOperator.serviceMonitor.spec.endpoints.tlsConfig]\n          insecureSkipVerify = true"
 }

bench/resources/refresh_golden_outputs.sh

@@ -4,13 +4,8 @@ set -euo pipefail
 ROOT_DIR="$(git rev-parse --show-toplevel)"
 pushd "$ROOT_DIR" || exit 1
 
-# Use the same Scala version the bench actually runs under (scalaVersions.head) so
-# goldens never drift from the runtime. Pinning a fixed version here previously broke
-# refreshes once that version stopped building.
-SCALA_VERSION="$(./mill show bench.scalaVersion 2>/dev/null | tr -d '"[:space:]')"
-echo "Using Scala $SCALA_VERSION (bench.scalaVersion)"
-./mill show "sjsonnet.jvm[$SCALA_VERSION].assembly"
-SJSONNET="out/sjsonnet/jvm/$SCALA_VERSION/assembly.dest/out.jar"
+./mill show "sjsonnet.jvm[3.3.7].assembly"
+SJSONNET="out/sjsonnet/jvm/3.3.7/assembly.dest/out.jar"
 
 for suite in bench/resources/*_suite; do
   suite_name=$(basename "$suite")

bench/resources/sjsonnet_suite/range_sum_avg.jsonnet.golden

@@ -1 +1 @@
-2000002000000.5
+2.0000020000005E12

sjsonnet/src-js/sjsonnet/Platform.scala

@@ -523,9 +523,6 @@ object Platform {
     throw new Exception("SHA3 not implemented in Scala.js")
   }
 
-  def hashBytes(bytes: Array[Byte]): String =
-    scala.util.hashing.MurmurHash3.bytesHash(bytes).toHexString
-
   private val regexCache = new util.concurrent.ConcurrentHashMap[String, Pattern]
   private val namedGroupPattern = Pattern.compile("\\(\\?<(.+?)>.*?\\)")
   private val namedGroupPatternReplace = Pattern.compile("(\\(\\?P<)(.+?>.*?\\))")

sjsonnet/src/sjsonnet/Importer.scala

@@ -186,55 +186,15 @@ final case class StaticResolvedFile(content: String) extends ResolvedFile {
   override def readRawBytes(): Array[Byte] = content.getBytes(StandardCharsets.UTF_8)
 }
 
-final class StaticBinaryResolvedFile(content0: Array[Byte])
-    extends ResolvedFile
-    with Product
-    with Serializable {
-  private val bytes: Array[Byte] = content0.clone()
-
+final case class StaticBinaryResolvedFile(content: Array[Byte]) extends ResolvedFile {
   def getParserInput(): ParserInput = throw new NotImplementedError("Not used for binary imports")
 
   def readString(): String = throw new NotImplementedError("Not used for binary imports")
 
-  def content: Array[Byte] = bytes
-
-  private lazy val contentHashValue: String = Platform.hashBytes(bytes)
-
-  def contentHash(): String = contentHashValue
-
-  override def readRawBytes(): Array[Byte] = bytes
-
-  def copy(content: Array[Byte] = this.content): StaticBinaryResolvedFile =
-    new StaticBinaryResolvedFile(content)
-
-  def productArity: Int = 1
-
-  def productElement(n: Int): Any =
-    if (n == 0) content else throw new IndexOutOfBoundsException(n.toString)
-
-  def canEqual(that: Any): Boolean = that.isInstanceOf[StaticBinaryResolvedFile]
-
-  override def productPrefix: String = "StaticBinaryResolvedFile"
-
-  override def equals(that: Any): Boolean = that match {
-    case other: StaticBinaryResolvedFile =>
-      (this eq other) || (other.canEqual(this) && java.util.Arrays.equals(bytes, other.bytes))
-    case _ => false
-  }
-
-  override def hashCode(): Int = java.util.Arrays.hashCode(bytes)
-
-  override def toString: String = productIterator.mkString(productPrefix + "(", ",", ")")
-}
-
-object StaticBinaryResolvedFile
-    extends scala.runtime.AbstractFunction1[Array[Byte], StaticBinaryResolvedFile]
-    with Serializable {
-  def apply(content: Array[Byte]): StaticBinaryResolvedFile =
-    new StaticBinaryResolvedFile(content)
+  // We just cheat, the content hash can be the content itself for static imports
+  def contentHash(): String = content.hashCode().toString
 
-  def unapply(file: StaticBinaryResolvedFile): Option[Array[Byte]] =
-    if (file == null) None else Some(file.content)
+  override def readRawBytes(): Array[Byte] = content
 }
 
 class CachedImporter(parent: Importer) extends Importer {

sjsonnet/src/sjsonnet/StaticOptimizer.scala

@@ -138,11 +138,11 @@ class StaticOptimizer(
       case e @ UnaryOp(pos, op, v: Val) => tryFoldUnaryOp(pos, op, v, e)
 
       // Branch elimination: constant condition in if-else
-      case IfElse(_, _: Val.True, thenExpr, _) =>
-        thenExpr
-      case IfElse(_, _: Val.False, _, elseExpr) =>
-        if (elseExpr == null) Val.staticNull
-        else elseExpr
+      case IfElse(pos, _: Val.True, thenExpr, _) =>
+        thenExpr.pos = pos; thenExpr
+      case IfElse(pos, _: Val.False, _, elseExpr) =>
+        if (elseExpr == null) Val.Null(pos)
+        else { elseExpr.pos = pos; elseExpr }
 
       // Short-circuit elimination for And/Or with constant lhs.
       //
@@ -152,12 +152,10 @@ class StaticOptimizer(
       // into just `rhs` without the Bool guard, we silently remove that runtime type check,
       // causing programs like `true && "hello"` to return "hello" instead of erroring.
       // See: Evaluator.visitAnd / Evaluator.visitOr for the authoritative runtime semantics.
-      // The lhs-result cases return lhs itself to keep the selected boolean's position and avoid
-      // turning the already-created literal/folded lhs into garbage.
-      case And(_, _: Val.True, rhs: Val.Bool) => rhs
-      case And(_, lhs: Val.False, _)          => lhs
-      case Or(_, lhs: Val.True, _)            => lhs
-      case Or(_, _: Val.False, rhs: Val.Bool) => rhs
+      case And(pos, _: Val.True, rhs: Val.Bool) => rhs.pos = pos; rhs
+      case And(pos, _: Val.False, _)            => Val.False(pos)
+      case Or(pos, _: Val.True, _)              => Val.True(pos)
+      case Or(pos, _: Val.False, rhs: Val.Bool) => rhs.pos = pos; rhs
 
       // Identity-equivalent function recognition. Cheap pattern match here keeps the runtime
       // fast path (`Val.Func.isEffectivelyIdentity`) at single-field cost.

sjsonnet/src/sjsonnet/Val.scala

@@ -1815,16 +1815,6 @@ object Val {
     /** Package-private ref to DebugStats, set by Evaluator when --debug-stats is active. */
     private[sjsonnet] var currentDebugStats: DebugStats = _
 
-    /**
-     * Field count at/below which `valueRaw` resolves an inline-array object by linear scan. Above
-     * it (e.g. large imported strict-JSON objects, whose inline arrays are uncapped — see
-     * Importer), `valueRaw` instead lazily builds the `value0` LinkedHashMap via `getValue0` for
-     * O(1) lookups. This keeps the scan for small objects (and never builds a map for the
-     * import-and-emit path, which materializes via direct inline iteration) while avoiding O(N^2)
-     * when a large imported object gains a `super` and must be resolved per-key.
-     */
-    private[sjsonnet] final val InlineScanMax = 8
-
     /**
      * @param add
      *   whether this field was defined the "+:", "+::" or "+:::" separators, corresponding to the
@@ -2489,10 +2479,8 @@ object Val {
           } else {
             if (s == null) null else s.valueRaw(k, self, pos, cacheOwner, cacheKey)
           }
-        } else if (inlineFieldKeys != null && inlineFieldKeys.length <= Obj.InlineScanMax) {
-          // Inline multi-field fast path: linear scan over small arrays. Large inline objects
-          // (uncapped imported strict-JSON) fall through to the getValue0 path below, which lazily
-          // builds an O(1) lookup map instead of scanning O(N) per key.
+        } else if (inlineFieldKeys != null) {
+          // Inline multi-field fast path: linear scan over small arrays
           val keys = inlineFieldKeys
           val members = inlineFieldMembers
           val n = keys.length

sjsonnet/src/sjsonnet/functions/FunctionModule.scala

@@ -34,11 +34,7 @@ trait FunctionModule extends FunctionBuilder {
    *   the module object
    */
   def moduleFromFunctions(functions: (String, Val.Func)*): Val.Obj = {
-    Val.Obj.mkWithConstCache(
-      dummyPos,
-      functions.size,
-      functions.map { case (k, v) => (k, memberOf(v)) }
-    )
+    Val.Obj.mk(dummyPos, functions.map { case (k, v) => (k, memberOf(v)) }: _*)
   }
 
   /**
@@ -50,11 +46,7 @@ trait FunctionModule extends FunctionBuilder {
    *   the module object
    */
   def moduleFromModules(subModules: FunctionModule*): Val.Obj = {
-    Val.Obj.mkWithConstCache(
-      dummyPos,
-      subModules.size,
-      subModules.map { module => (module.name, memberOf(module.module)) }
-    )
+    Val.Obj.mk(dummyPos, subModules.map { module => (module.name, memberOf(module.module)) }: _*)
   }
 
   /**

sjsonnet/src/sjsonnet/stdlib/ArrayModule.scala

@@ -616,11 +616,6 @@ object ArrayModule extends AbstractFunctionModule {
   }
 
   private object Range extends Val.Builtin2("range", "from", "to") {
-    // Not static-safe: constant folding would call asStrictArray on the result, eagerly
-    // materializing the lazy RangeArr at optimization time. That defeats the lazy RangeArr /
-    // ConcatView / sharedConcatPrefixLength fast path (e.g. `std.range(1, N) + [x] < ... + [y]`
-    // collapses from O(1) to O(N) materialization on every evaluation).
-    override def staticSafe: Boolean = false
     def evalRhs(from: Eval, to: Eval, ev: EvalScope, pos: Position): Val = {
       val fromInt = requireInt(from.value, "from", pos)(ev)
       val toInt = requireInt(to.value, "to", pos)(ev)

sjsonnet/src/sjsonnet/stdlib/ManifestModule.scala

@@ -130,6 +130,8 @@ object ManifestModule extends AbstractFunctionModule {
       } catch {
         case e: ujson.ParseException =>
           throw Error.fail("Invalid JSON: " + e.getMessage, pos)(ev)
+        case _: ujson.IncompleteParseException =>
+          throw Error.fail("Invalid JSON: unexpected end of JSON input", pos)(ev)
       }
     }
   }