Add importbin to sjsonnet (#251)

Per spec
https://jsonnet.org/ref/spec.html#:~:text=UTF%2D8.%20For-,importbin,-%2C%20the%20file%20is

importbin imports the file as an array of ints from 0-255

bench/src/main/scala/sjsonnet/MainBenchmark.scala

@@ -31,7 +31,7 @@ object MainBenchmark {
       parseCache = parseCache
     )
     val renderer = new Renderer(new StringWriter, indent = 3)
-    interp.interpret0(interp.resolver.read(path).get.readString(), path, renderer).getOrElse(???)
+    interp.interpret0(interp.resolver.read(path, binaryData = false).get.readString(), path, renderer).getOrElse(???)
     (parseCache.keySet.toIndexedSeq, interp.evaluator)
   }
 

bench/src/main/scala/sjsonnet/RunProfiler.scala

@@ -24,7 +24,7 @@ object RunProfiler extends App {
 
   def run(): Long = {
     val renderer = new Renderer(new StringWriter, indent = 3)
-    val start = interp.resolver.read(path).get.readString()
+    val start = interp.resolver.read(path, binaryData = false).get.readString()
     val t0 = System.nanoTime()
     interp.interpret0(start, path, renderer).getOrElse(???)
     System.nanoTime() - t0

sjsonnet/src-js/sjsonnet/SjsonnetMain.scala

@@ -24,7 +24,7 @@ object SjsonnetMain {
             case null => None
             case s => Some(JsVirtualPath(s))
           }
-        def read(path: Path): Option[ResolvedFile] =
+        def read(path: Path, binaryData: Boolean): Option[ResolvedFile] =
           Option(StaticResolvedFile(importLoader(path.asInstanceOf[JsVirtualPath].path)))
       },
       parseCache = new DefaultParseCache,

sjsonnet/src-jvm-native/sjsonnet/CachedResolvedFile.scala

@@ -1,11 +1,11 @@
 package sjsonnet
 
-import java.io.{BufferedInputStream, File, FileInputStream}
+import fastparse.ParserInput
+
+import java.io.File
 import java.nio.charset.StandardCharsets
 import java.nio.file.Files
 
-import fastparse.ParserInput
-
 /**
  * A class that encapsulates a resolved import. This is used to cache the result of
  * resolving an import. If the import is deemed too large (IE it's a large file), then we will avoid keeping it in
@@ -18,16 +18,17 @@ import fastparse.ParserInput
  * @param cacheThresholdBytes The maximum size of a file that we will cache in memory. If the file
  * is larger than this, then we will serve it from disk
  */
-class CachedResolvedFile(val resolvedImportPath: OsPath, memoryLimitBytes: Long, cacheThresholdBytes: Long = 1024 * 1024) extends ResolvedFile {
+class CachedResolvedFile(val resolvedImportPath: OsPath, memoryLimitBytes: Long, cacheThresholdBytes: Long = 1024 * 1024, binaryData: Boolean = false) extends ResolvedFile {
 
   private val jFile: File = resolvedImportPath.p.toIO
 
-  assert(jFile.exists(), s"Resolved import path ${resolvedImportPath} does not exist")
+  assert(jFile.exists(), s"Resolved import path $resolvedImportPath does not exist")
   // Assert that the file is less than limit
-  assert(jFile.length() <= memoryLimitBytes, s"Resolved import path ${resolvedImportPath} is too large: ${jFile.length()} bytes > ${memoryLimitBytes} bytes")
+  assert(jFile.length() <= memoryLimitBytes, s"Resolved import path $resolvedImportPath is too large: ${jFile.length()} bytes > ${memoryLimitBytes} bytes")
 
   private[this] val resolvedImportContent: StaticResolvedFile = {
-    if (jFile.length() > cacheThresholdBytes) {
+    // TODO: Support caching binary data
+    if (jFile.length() > cacheThresholdBytes || binaryData) {
       // If the file is too large, then we will just read it from disk
       null
     } else {
@@ -39,6 +40,8 @@ class CachedResolvedFile(val resolvedImportPath: OsPath, memoryLimitBytes: Long,
     new String(Files.readAllBytes(jFile.toPath), StandardCharsets.UTF_8);
   }
 
+  private[this] def readRawBytes(jFile: File): Array[Byte] = Files.readAllBytes(jFile.toPath)
+
   /**
    * A method that will return a reader for the resolved import. If the import is too large, then this will return
    * a reader that will read the file from disk. Otherwise, it will return a reader that reads from memory.
@@ -61,6 +64,7 @@ class CachedResolvedFile(val resolvedImportPath: OsPath, memoryLimitBytes: Long,
     }
   }
 
+
   override lazy val contentHash: String = {
     if (resolvedImportContent == null) {
       // If the file is too large, then we will just read it from disk
@@ -69,5 +73,15 @@ class CachedResolvedFile(val resolvedImportPath: OsPath, memoryLimitBytes: Long,
       resolvedImportContent.contentHash
     }
   }
+
+  override def readRawBytes(): Array[Byte] = {
+    if (resolvedImportContent == null) {
+      // If the file is too large, then we will just read it from disk
+      readRawBytes(jFile)
+    } else {
+      // Otherwise, we will read it from memory
+      resolvedImportContent.readRawBytes()
+    }
+  }
 }
 

sjsonnet/src-jvm-native/sjsonnet/SjsonnetMain.scala

@@ -22,8 +22,8 @@ object SjsonnetMain {
         .find(os.exists)
         .flatMap(p => try Some(OsPath(p)) catch{case NonFatal(_) => None})
 
-    def read(path: Path): Option[ResolvedFile] = {
-      readPath(path)
+    def read(path: Path, binaryData: Boolean = false): Option[ResolvedFile] = {
+      readPath(path, binaryData)
     }
   }
 
@@ -200,8 +200,8 @@ object SjsonnetMain {
         case Some(i) => new Importer {
           def resolve(docBase: Path, importName: String): Option[Path] =
             i(docBase, importName).map(OsPath)
-          def read(path: Path): Option[ResolvedFile] = {
-            readPath(path)
+          def read(path: Path, binaryData: Boolean): Option[ResolvedFile] = {
+            readPath(path, binaryData)
           }
         }
         case None => resolveImport(config.jpaths.map(os.Path(_, wd)).map(OsPath), allowedInputs)
@@ -295,10 +295,10 @@ object SjsonnetMain {
    * of caching on top of the underlying file system. Small files are read into memory, while large
    * files are read from disk.
    */
-  private[this] def readPath(path: Path): Option[ResolvedFile] = {
+  private[this] def readPath(path: Path, binaryData: Boolean): Option[ResolvedFile] = {
     val osPath = path.asInstanceOf[OsPath].p
     if (os.exists(osPath) && os.isFile(osPath)) {
-      Some(new CachedResolvedFile(path.asInstanceOf[OsPath], memoryLimitBytes = Int.MaxValue.toLong))
+      Some(new CachedResolvedFile(path.asInstanceOf[OsPath], memoryLimitBytes = Int.MaxValue.toLong, binaryData = binaryData))
     } else {
       None
     }

sjsonnet/src/sjsonnet/Error.scala

@@ -101,7 +101,7 @@ trait EvalErrorScope {
   def wd: Path
 
   def prettyIndex(pos: Position): Option[(Int, Int)] = {
-    importer.read(pos.currentFile).map { s =>
+    importer.read(pos.currentFile, binaryData = false).map { s =>
       val Array(line, col) =
         s.getParserInput().prettyIndex(pos.offset).split(':')
       (line.toInt, col.toInt)

sjsonnet/src/sjsonnet/Evaluator.scala

@@ -61,6 +61,7 @@ class Evaluator(resolver: CachedResolver,
       case e: Import => visitImport(e)
       case e: Apply0 => visitApply0(e)
       case e: ImportStr => visitImportStr(e)
+      case e: ImportBin => visitImportBin(e)
       case e: Expr.Error => visitError(e)
       case e => visitInvalid(e)
     }
@@ -301,10 +302,14 @@ class Evaluator(resolver: CachedResolver,
   }
 
   def visitImportStr(e: ImportStr)(implicit scope: ValScope): Val.Str =
-    Val.Str(e.pos, importer.resolveAndReadOrFail(e.value, e.pos)._2.readString())
+    Val.Str(e.pos, importer.resolveAndReadOrFail(e.value, e.pos, binaryData = false)._2.readString())
+
+  def visitImportBin(e: ImportBin): Val.Arr =
+    new Val.Arr(e.pos, importer.resolveAndReadOrFail(e.value, e.pos, binaryData = true)._2.readRawBytes().map(
+      x => Val.Num(e.pos, (x & 0xff).doubleValue)))
 
   def visitImport(e: Import)(implicit scope: ValScope): Val = {
-    val (p, str) = importer.resolveAndReadOrFail(e.value, e.pos)
+    val (p, str) = importer.resolveAndReadOrFail(e.value, e.pos, binaryData = false)
     cachedImports.getOrElseUpdate(
       p,
       {

sjsonnet/src/sjsonnet/Expr.scala

@@ -126,6 +126,7 @@ object Expr{
   case class Bind(pos: Position, name: String, args: Params, rhs: Expr) extends Member
   case class Import(pos: Position, value: String) extends Expr
   case class ImportStr(pos: Position, value: String) extends Expr
+  case class ImportBin(pos: Position, value: String) extends Expr
   case class Error(pos: Position, value: Expr) extends Expr
   case class Apply(pos: Position, value: Expr, args: Array[Expr], namedNames: Array[String]) extends Expr
   case class Apply0(pos: Position, value: Expr) extends Expr
@@ -181,5 +182,4 @@ object Expr{
       override def toString = s"ObjComp($pos, ${arrStr(preLocals)}, $key, $value, ${arrStr(postLocals)}, $first, $rest)"
     }
   }
-
 }

sjsonnet/src/sjsonnet/Importer.scala

@@ -1,33 +1,31 @@
 package sjsonnet
 
-import java.io.{BufferedInputStream, BufferedReader, ByteArrayInputStream, File, FileInputStream, FileReader, InputStream, RandomAccessFile, Reader, StringReader}
-import java.nio.file.Files
-import java.security.MessageDigest
-import scala.collection.mutable
 import fastparse.{IndexedParserInput, Parsed, ParserInput}
 
+import java.io.{BufferedInputStream, File, FileInputStream, RandomAccessFile}
 import java.nio.charset.StandardCharsets
+import scala.collection.mutable
 
 
 /** Resolve and read imported files */
 abstract class Importer {
   def resolve(docBase: Path, importName: String): Option[Path]
-  def read(path: Path): Option[ResolvedFile]
+  def read(path: Path, binaryData: Boolean): Option[ResolvedFile]
 
-  def resolveAndRead(docBase: Path, importName: String): Option[(Path, ResolvedFile)] = for {
+  private def resolveAndRead(docBase: Path, importName: String, binaryData: Boolean): Option[(Path, ResolvedFile)] = for {
     path <- resolve(docBase, importName)
-    txt <- read(path)
+    txt <- read(path, binaryData)
   } yield (path, txt)
 
-  def resolveAndReadOrFail(value: String, pos: Position)(implicit ev: EvalErrorScope): (Path, ResolvedFile) =
-    resolveAndRead(pos.fileScope.currentFile.parent(), value)
+  def resolveAndReadOrFail(value: String, pos: Position, binaryData: Boolean)(implicit ev: EvalErrorScope): (Path, ResolvedFile) =
+    resolveAndRead(pos.fileScope.currentFile.parent(), value, binaryData = binaryData)
       .getOrElse(Error.fail("Couldn't import file: " + pprint.Util.literalize(value), pos))
 }
 
 object Importer {
   val empty: Importer = new Importer {
     def resolve(docBase: Path, importName: String): Option[Path] = None
-    def read(path: Path): Option[ResolvedFile] = None
+    def read(path: Path, binaryData: Boolean): Option[ResolvedFile] = None
   }
 }
 
@@ -97,7 +95,7 @@ class BufferedRandomAccessFile(fileName: String, bufferSize: Int) {
   private val fileLength: Long = file.length()
 
   private def fillBuffer(position: Long): Unit = {
-    if (file.getFilePointer() != position) {
+    if (file.getFilePointer != position) {
       file.seek(position)
     }
     val bytesRead = file.read(buffer, 0, bufferSize)
@@ -150,6 +148,9 @@ trait ResolvedFile {
 
   // Get a content hash of the file suitable for detecting changes in a given file.
   def contentHash(): String
+
+  // Used by importbin
+  def readRawBytes(): Array[Byte]
 }
 
 case class StaticResolvedFile(content: String) extends ResolvedFile {
@@ -159,19 +160,21 @@ case class StaticResolvedFile(content: String) extends ResolvedFile {
 
   // We just cheat, the content hash can be the content itself for static imports
   lazy val contentHash: String = content
+
+  override def readRawBytes(): Array[Byte] = content.getBytes(StandardCharsets.UTF_8)
 }
 
 class CachedImporter(parent: Importer) extends Importer {
   val cache = mutable.HashMap.empty[Path, ResolvedFile]
 
   def resolve(docBase: Path, importName: String): Option[Path] = parent.resolve(docBase, importName)
 
-  def read(path: Path): Option[ResolvedFile] = cache.get(path) match {
+  def read(path: Path, binaryData: Boolean): Option[ResolvedFile] = cache.get(path) match {
     case s @ Some(x) =>
       if(x == null) None else s
     case None =>
-      val x = parent.read(path)
-      cache.put(path, x.getOrElse(null))
+      val x = parent.read(path, binaryData)
+      cache.put(path, x.orNull)
       x
   }
 }
@@ -184,7 +187,7 @@ class CachedResolver(
   internedStaticFieldSets: mutable.HashMap[Val.StaticObjectFieldSet, java.util.LinkedHashMap[String, java.lang.Boolean]]) extends CachedImporter(parentImporter) {
 
   def parse(path: Path, content: ResolvedFile)(implicit ev: EvalErrorScope): Either[Error, (Expr, FileScope)] = {
-    parseCache.getOrElseUpdate((path, content.contentHash.toString), {
+    parseCache.getOrElseUpdate((path, content.contentHash()), {
       val parsed = fastparse.parse(content.getParserInput(), new Parser(path, strictImportSyntax, internedStrings, internedStaticFieldSets).document(_)) match {
         case f @ Parsed.Failure(_, _, _) =>
           val traced = f.trace()

sjsonnet/src/sjsonnet/Parser.scala

@@ -36,7 +36,7 @@ object Parser {
 
   val keywords = Set(
     "assert", "else", "error", "false", "for", "function", "if", "import", "importstr",
-    "in", "local", "null", "tailstrict", "then", "self", "super", "true"
+    "in", "local", "null", "tailstrict", "then", "self", "super", "true", "importbin"
   )
 
   def idStartChar(c: Char) = c == '_' || ('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z')
@@ -239,6 +239,7 @@ class Parser(val currentFile: Path,
 
   def local[_: P] = P( localExpr )
   def importStr[_: P](pos: Position) = P( importExpr.map(Expr.ImportStr(pos, _)) )
+  def importBin[_: P](pos: Position) = P( importExpr.map(Expr.ImportBin(pos, _)) )
   def `import`[_: P](pos: Position) = P( importExpr.map(Expr.Import(pos, _)) )
   def error[_: P](pos: Position) = P(expr.map(Expr.Error(pos, _)) )
 
@@ -298,6 +299,7 @@ class Parser(val currentFile: Path,
               case "if"        => Pass ~ ifElse(pos)
               case "function"  => Pass ~ function(pos)
               case "importstr" => Pass ~ importStr(pos)
+              case "importbin" => Pass ~ importBin(pos)
               case "import"    => Pass ~ `import`(pos)
               case "error"     => Pass ~ error(pos)
               case "assert"    => Pass ~ assertExpr(pos)

sjsonnet/test/resources/test_suite/import.jsonnet

@@ -24,14 +24,17 @@ std.assertEqual((import 'lib/A_20_func.libsonnet')(), 20) &&
 std.assertEqual((import 'lib/A_20_func.libsonnet')(), 20) &&
 // The block string is hard to test because the filename would include a terminating \n
 
-// Each import has its own environment, can't be overidden.
+// Each import has its own environment, can't be overridden.
 std.assertEqual(local A = 7; local lib = import 'lib/A_20.libsonnet'; lib, 20) &&
 std.assertEqual(local A = 7, lib = import 'lib/A_20.libsonnet'; lib, 20) &&
 
 std.assertEqual(importstr 'lib/some_file.txt', 'Hello World!\n') &&
 std.assertEqual(importstr 'lib/some_file.txt', 'Hello World!\n') &&
 
+std.assertEqual(importbin 'lib/nonutf8.bin', [255, 0, 254]) &&
+std.assertEqual(importbin 'lib/nonutf8.bin', [255, 0, 254]) &&
+
 std.assertEqual(import 'lib/rel_path.libsonnet', 'rel_path') &&
 std.assertEqual(import 'lib/rel_path4.libsonnet', 'rel_path') &&
 
-true
+true

sjsonnet/test/src/sjsonnet/Std0150FunctionsTests.scala

@@ -89,7 +89,7 @@ object Std0150FunctionsTests extends TestSuite {
         parseCache = new DefaultParseCache,
       )
 
-      def check(s: String, expected: ujson.Value) =
+      def check(s: String, expected: ujson.Value): Unit =
         interpreter.interpret(s, DummyPath("(memory)")) ==> Right(expected)
 
       check("""std.extVar("num")""", 1)
@@ -121,14 +121,14 @@ object Std0150FunctionsTests extends TestSuite {
           override def resolve(docBase: Path, importName: String): Option[Path] = importName match{
             case "bar.json" => Some(DummyPath("bar"))
           }
-          override def read(path: Path): Option[ResolvedFile] = path match{
+          override def read(path: Path, binaryData: Boolean): Option[ResolvedFile] = path match{
             case DummyPath("bar") => Some(StaticResolvedFile("""{"x": "y"}"""))
           }
         },
         parseCache = new DefaultParseCache,
       )
 
-      def check(s: String, expected: ujson.Value) =
+      def check(s: String, expected: ujson.Value): Unit =
         interpreter.interpret(s, DummyPath("(memory)")) ==> Right(expected)
 
       check("""function(num) num""", 1)