STDREGEX

build.sbt

@@ -21,6 +21,7 @@ lazy val main = (project in file("sjsonnet"))
       "org.scala-lang.modules" %% "scala-collection-compat" % "2.11.0",
       "org.tukaani" % "xz" % "1.8",
       "org.yaml" % "snakeyaml" % "1.33",
+      "com.google.re2j" % "re2j" % "1.7",
     ),
     libraryDependencies ++= Seq(
       "com.lihaoyi" %% "utest" % "0.8.2",

build.sc

@@ -108,7 +108,8 @@ object sjsonnet extends Module {
       ivy"org.json:json:20240303",
       ivy"org.tukaani:xz::1.10",
       ivy"org.lz4:lz4-java::1.8.0",
-      ivy"org.yaml:snakeyaml::1.33"
+      ivy"org.yaml:snakeyaml::1.33",
+      ivy"com.google.re2j:re2j:1.7",
     )
     def scalacOptions = Seq("-opt:l:inline", "-opt-inline-from:sjsonnet.**")
 

sjsonnet/src-js/sjsonnet/Platform.scala

@@ -1,5 +1,10 @@
 package sjsonnet
+
 import java.io.File
+import java.util
+import java.util.regex.Pattern
+
+
 object Platform {
   def gzipBytes(s: Array[Byte]): String = {
     throw new Exception("GZip not implemented in Scala.js")
@@ -34,4 +39,11 @@ object Platform {
   def hashFile(file: File): String = {
     throw new Exception("hashFile not implemented in Scala.js")
   }
+
+  private val regexCache = new util.LinkedHashMap[String, Pattern](100, 0.75f, true) {
+    override def removeEldestEntry(eldest: util.Map.Entry[String, Pattern]): Boolean = size() > 100
+  }
+  def getPatternFromCache(pat: String) : Pattern = regexCache.computeIfAbsent(pat, _ => Pattern.compile(pat))
+
+  def regexQuote(s: String): String = Pattern.quote(s)
 }

sjsonnet/src-jvm/sjsonnet/Platform.scala

@@ -1,14 +1,17 @@
 package sjsonnet
 
 import java.io.{BufferedInputStream, ByteArrayOutputStream, File, FileInputStream}
+import java.util
 import java.util.Base64
 import java.util.zip.GZIPOutputStream
+import com.google.re2j.Pattern
 import net.jpountz.xxhash.{StreamingXXHash64, XXHashFactory}
 import org.json.{JSONArray, JSONObject}
 import org.tukaani.xz.LZMA2Options
 import org.tukaani.xz.XZOutputStream
 import org.yaml.snakeyaml.{LoaderOptions, Yaml}
 import org.yaml.snakeyaml.constructor.SafeConstructor
+
 import scala.jdk.CollectionConverters._
 
 object Platform {
@@ -107,4 +110,11 @@ object Platform {
 
     hash.getValue.toString
   }
+
+  private val regexCache = new util.LinkedHashMap[String, Pattern](100, 0.75f, true) {
+    override def removeEldestEntry(eldest: util.Map.Entry[String, Pattern]): Boolean = size() > 100
+  }
+  def getPatternFromCache(pat: String) : Pattern = regexCache.computeIfAbsent(pat, _ => Pattern.compile(pat))
+
+  def regexQuote(s: String): String = Pattern.quote(s)
 }

sjsonnet/src-native/sjsonnet/Platform.scala

@@ -1,8 +1,10 @@
 package sjsonnet
 
 import java.io.{ByteArrayOutputStream, File}
+import java.util
 import java.util.Base64
 import java.util.zip.GZIPOutputStream
+import java.util.regex.Pattern
 
 object Platform {
   def gzipBytes(b: Array[Byte]): String = {
@@ -50,4 +52,11 @@ object Platform {
     // File hashes in Scala Native are just the file content
     scala.io.Source.fromFile(file).mkString
   }
+
+  private val regexCache = new util.LinkedHashMap[String, Pattern](100, 0.75f, true) {
+    override def removeEldestEntry(eldest: util.Map.Entry[String, Pattern]): Boolean = size() > 100
+  }
+  def getPatternFromCache(pat: String) : Pattern = regexCache.computeIfAbsent(pat, _ => Pattern.compile(pat))
+
+  def regexQuote(s: String): String = Pattern.quote(s)
 }

sjsonnet/src/sjsonnet/PrettyYamlRenderer.scala

@@ -1,10 +1,8 @@
 package sjsonnet
 
 import java.io.{StringWriter, Writer}
-import java.util.regex.Pattern
 
 import upickle.core.{ArrVisitor, ObjVisitor}
-import fastparse.IndexedParserInput
 
 import scala.collection.mutable
 /**
@@ -240,7 +238,7 @@ object PrettyYamlRenderer{
    */
   def writeBlockString(str: String, out: Writer, depth: Int, indent: Int, lineComment: String) = {
     val len = str.length()
-    val splits = YamlRenderer.newlinePattern.split(str, -1)
+    val splits = Platform.getPatternFromCache("\n").split(str, -1)
     val blockOffsetNumeral = if (str.charAt(0) != ' ') "" else indent
     val (blockStyle, dropRight) =
       (str.charAt(len - 1), if (len > 2) Some(str.charAt(len - 2)) else None) match{

sjsonnet/src/sjsonnet/Std.scala

@@ -4,12 +4,10 @@ import java.io.StringWriter
 import java.nio.charset.StandardCharsets.UTF_8
 import java.util.Base64
 import java.util
-import java.util.regex.Pattern
 import sjsonnet.Expr.Member.Visibility
 
 import scala.collection.Searching._
 import scala.collection.mutable
-import scala.util.matching.Regex
 
 /**
   * The Jsonnet standard library, `std`, with each builtin function implemented
@@ -19,8 +17,8 @@ import scala.util.matching.Regex
 class Std(private val additionalNativeFunctions: Map[String, Val.Builtin] = Map.empty) {
   private val dummyPos: Position = new Position(null, 0)
   private val emptyLazyArray = new Array[Lazy](0)
-  private val leadingWhiteSpacePattern = Pattern.compile("^[ \t\n\f\r\u0085\u00A0']+")
-  private val trailingWhiteSpacePattern = Pattern.compile("[ \t\n\f\r\u0085\u00A0']+$")
+  private val leadingWhiteSpacePattern = Platform.getPatternFromCache("^[ \t\n\f\r\u0085\u00A0']+")
+  private val trailingWhiteSpacePattern = Platform.getPatternFromCache("[ \t\n\f\r\u0085\u00A0']+$")
   private val oldNativeFunctions = Map(
     builtin("gzip", "v"){ (_, _, v: Val) =>
       v match{
@@ -48,7 +46,7 @@ class Std(private val additionalNativeFunctions: Map[String, Val.Builtin] = Map.
     },
   )
   require(oldNativeFunctions.forall(k => !additionalNativeFunctions.contains(k._1)), "Conflicting native functions")
-  private val nativeFunctions = oldNativeFunctions ++ additionalNativeFunctions
+  private val nativeFunctions = oldNativeFunctions ++ additionalNativeFunctions ++ StdRegex.functions
 
   private object AssertEqual extends Val.Builtin2("assertEqual", "a", "b") {
     def evalRhs(v1: Val, v2: Val, ev: EvalScope, pos: Position): Val = {
@@ -474,26 +472,24 @@ class Std(private val additionalNativeFunctions: Map[String, Val.Builtin] = Map.
       Val.Str(pos, str.asString.replaceAll(from.asString, to.asString))
     override def specialize(args: Array[Expr]) = args match {
       case Array(str, from: Val.Str, to) =>
-        try { (new SpecFrom(Pattern.compile(from.value)), Array(str, to)) } catch { case _: Exception => null }
+        try { (new SpecFrom(from.value), Array(str, to)) } catch { case _: Exception => null }
       case _ => null
     }
-    private class SpecFrom(from: Pattern) extends Val.Builtin2("strReplaceAll", "str", "to") {
+    private class SpecFrom(from: String) extends Val.Builtin2("strReplaceAll", "str", "to") {
       def evalRhs(str: Val, to: Val, ev: EvalScope, pos: Position): Val =
-        Val.Str(pos, from.matcher(str.asString).replaceAll(to.asString))
+        Val.Str(pos, Platform.getPatternFromCache(from).matcher(str.asString).replaceAll(to.asString))
     }
   }
 
   private object StripUtils {
-    private def getLeadingPattern(chars: String): Pattern =
-      Pattern.compile("^[" + Regex.quote(chars) + "]+")
+    private def getLeadingPattern(chars: String): String = "^[" + Platform.regexQuote(chars) + "]+"
 
-    private def getTrailingPattern(chars: String): Pattern =
-      Pattern.compile("[" + Regex.quote(chars) + "]+$")
+    private def getTrailingPattern(chars: String): String = "[" + Platform.regexQuote(chars) + "]+$"
 
     def unspecializedStrip(str: String, chars: String, left: Boolean, right: Boolean): String = {
       var s = str
-      if (right) s = getTrailingPattern(chars).matcher(s).replaceAll("")
-      if (left) s = getLeadingPattern(chars).matcher(s).replaceAll("")
+      if (right) s = Platform.getPatternFromCache(getTrailingPattern(chars)).matcher(s).replaceAll("")
+      if (left) s = Platform.getPatternFromCache(getLeadingPattern(chars)).matcher(s).replaceAll("")
       s
     }
 
@@ -508,8 +504,8 @@ class Std(private val additionalNativeFunctions: Map[String, Val.Builtin] = Map.
 
       def evalRhs(str: Val, ev: EvalScope, pos: Position): Val = {
         var s = str.asString
-        if (right) s = rightPattern.matcher(s).replaceAll("")
-        if (left) s = leftPattern.matcher(s).replaceAll("")
+        if (right) s = Platform.getPatternFromCache(rightPattern).matcher(s).replaceAll("")
+        if (left) s = Platform.getPatternFromCache(leftPattern).matcher(s).replaceAll("")
         Val.Str(pos, s)
       }
     }

sjsonnet/src/sjsonnet/StdRegex.scala

@@ -0,0 +1,73 @@
+package sjsonnet
+
+import sjsonnet.Expr.Member.Visibility
+import sjsonnet.Val.Obj
+
+object StdRegex {
+  def functions: Map[String, Val.Builtin] = Map(
+    "regexPartialMatch" -> new Val.Builtin2("regexPartialMatch", "pattern", "str") {
+      override def evalRhs(pattern: Val, str: Val, ev: EvalScope, pos: Position): Val = {
+        val compiledPattern = Platform.getPatternFromCache(pattern.asString)
+        val matcher = compiledPattern.matcher(str.asString)
+        var returnStr: Val = Val.Null(pos.noOffset)
+        val captures = Array.newBuilder[Val.Str]
+        val groupCount = matcher.groupCount()
+        while (matcher.find()) {
+          if (returnStr.isInstanceOf[Val.Null]) {
+            returnStr = Val.Str(pos.noOffset, matcher.group(0))
+          }
+          for (i <- 1 to groupCount) {
+            captures += Val.Str(pos.noOffset, matcher.group(i))
+          }
+        }
+        val result = captures.result()
+        Val.Obj.mk(pos.noOffset,
+          "string" -> new Obj.ConstMember(true, Visibility.Normal, returnStr),
+          "captures" -> new Obj.ConstMember(true, Visibility.Normal, new Val.Arr(pos.noOffset, result))
+        )
+      }
+    },
+    "regexFullMatch" -> new Val.Builtin2("regexFullMatch", "pattern", "str") {
+      override def evalRhs(pattern: Val, str: Val, ev: EvalScope, pos: Position): Val = {
+        val compiledPattern = Platform.getPatternFromCache(pattern.asString)
+        val matcher = compiledPattern.matcher(str.asString)
+        if (!matcher.matches()) {
+          Val.Obj.mk(pos.noOffset,
+            "string" -> new Obj.ConstMember(true, Visibility.Normal, Val.Null(pos.noOffset)),
+            "captures" -> new Obj.ConstMember(true, Visibility.Normal, new Val.Arr(pos.noOffset, Array.empty[Lazy]))
+          )
+        } else {
+          val captures = Array.newBuilder[Val.Str]
+          val groupCount = matcher.groupCount()
+          for (i <- 0 to groupCount) {
+            captures += Val.Str(pos.noOffset, matcher.group(i))
+          }
+          val result = captures.result()
+          Val.Obj.mk(pos.noOffset,
+            "string" -> new Obj.ConstMember(true, Visibility.Normal, result.head),
+            "captures" -> new Obj.ConstMember(true, Visibility.Normal, new Val.Arr(pos.noOffset, result.drop(1)))
+          )
+        }
+      }
+    },
+    "regexGlobalReplace" -> new Val.Builtin3("regexGlobalReplace", "str", "pattern", "to") {
+      override def evalRhs(str: Val, pattern: Val, to: Val, ev: EvalScope, pos: Position): Val = {
+        val compiledPattern = Platform.getPatternFromCache(pattern.asString)
+        val matcher = compiledPattern.matcher(str.asString)
+        Val.Str(pos.noOffset, matcher.replaceAll(to.asString))
+      }
+    },
+    "regexReplace" -> new Val.Builtin3("regexGlobalReplace", "str", "pattern", "to") {
+      override def evalRhs(str: Val, pattern: Val, to: Val, ev: EvalScope, pos: Position): Val = {
+        val compiledPattern = Platform.getPatternFromCache(pattern.asString)
+        val matcher = compiledPattern.matcher(str.asString)
+        Val.Str(pos.noOffset, matcher.replaceFirst(to.asString))
+      }
+    },
+    "regexQuoteMeta" -> new Val.Builtin1("regexQuoteMeta", "str") {
+      override def evalRhs(str: Val, ev: EvalScope, pos: Position): Val = {
+        Val.Str(pos.noOffset, Platform.regexQuote(str.asString))
+      }
+    }
+  )
+}

sjsonnet/src/sjsonnet/TomlRenderer.scala

@@ -3,7 +3,6 @@ package sjsonnet
 import upickle.core.{ArrVisitor, CharBuilder, ObjVisitor, SimpleVisitor, Visitor}
 
 import java.io.StringWriter
-import java.util.regex.Pattern
 
 
 class TomlRenderer(out: StringWriter = new java.io.StringWriter(), cumulatedIndent: String, indent: String) extends SimpleVisitor[StringWriter, StringWriter]{
@@ -117,7 +116,7 @@ class TomlRenderer(out: StringWriter = new java.io.StringWriter(), cumulatedInde
 }
 
 object TomlRenderer {
-  private val bareAllowed = Pattern.compile("[A-Za-z0-9_-]+")
+  private val bareAllowed = Platform.getPatternFromCache("[A-Za-z0-9_-]+")
   def escapeKey(key: String): String = if (bareAllowed.matcher(key).matches()) key else {
     val out = new StringWriter()
     BaseRenderer.escape(out, key, unicode = true)

sjsonnet/src/sjsonnet/YamlRenderer.scala

@@ -1,12 +1,8 @@
 package sjsonnet
 
 import java.io.StringWriter
-import java.util.regex.Pattern
 import upickle.core.{ArrVisitor, ObjVisitor, SimpleVisitor, Visitor}
 
-import scala.util.Try
-
-
 
 class YamlRenderer(_out: StringWriter = new java.io.StringWriter(), indentArrayInObject: Boolean = false,
                    quoteKeys: Boolean = true, indent: Int = 2) extends BaseCharRenderer(_out, indent){
@@ -52,7 +48,7 @@ class YamlRenderer(_out: StringWriter = new java.io.StringWriter(), indentArrayI
       elemBuilder.append('"')
       elemBuilder.append('"')
     } else if (s.charAt(len - 1) == '\n') {
-      val splits = YamlRenderer.newlinePattern.split(s)
+      val splits = Platform.getPatternFromCache("\n").split(s.toString)
       elemBuilder.append('|')
       depth += 1
       splits.foreach { split =>
@@ -174,15 +170,14 @@ class YamlRenderer(_out: StringWriter = new java.io.StringWriter(), indentArrayI
   }
 }
 object YamlRenderer{
-  val newlinePattern: Pattern = Pattern.compile("\n")
-  private val safeYamlKeyPattern = Pattern.compile("^[a-zA-Z0-9/._-]+$")
+  private val safeYamlKeyPattern = Platform.getPatternFromCache("^[a-zA-Z0-9/._-]+$")
   private val yamlReserved = Set("true", "false", "null", "yes", "no", "on", "off", "y", "n", ".nan",
     "+.inf", "-.inf", ".inf", "null", "-", "---", "''")
-  private val yamlTimestampPattern = Pattern.compile("^(?:[0-9]*-){2}[0-9]*$")
-  private val yamlBinaryPattern = Pattern.compile("^[-+]?0b[0-1_]+$")
-  private val yamlHexPattern = Pattern.compile("[-+]?0x[0-9a-fA-F_]+")
-  private val yamlFloatPattern = Pattern.compile( "^-?([0-9_]*)*(\\.[0-9_]*)?(e[-+][0-9_]+)?$" )
-  private val yamlIntPattern = Pattern.compile("^[-+]?[0-9_]+$")
+  private val yamlTimestampPattern = Platform.getPatternFromCache("^(?:[0-9]*-){2}[0-9]*$")
+  private val yamlBinaryPattern = Platform.getPatternFromCache("^[-+]?0b[0-1_]+$")
+  private val yamlHexPattern = Platform.getPatternFromCache("[-+]?0x[0-9a-fA-F_]+")
+  private val yamlFloatPattern = Platform.getPatternFromCache( "^-?([0-9_]*)*(\\.[0-9_]*)?(e[-+][0-9_]+)?$" )
+  private val yamlIntPattern = Platform.getPatternFromCache("^[-+]?[0-9_]+$")
 
   private def isSafeBareKey(k: String) = {
     val l = k.toLowerCase

sjsonnet/test/src/sjsonnet/OldYamlRenderer.scala

@@ -20,7 +20,7 @@ class OldYamlRenderer(out: StringWriter = new java.io.StringWriter(), indentArra
     val len = s.length()
     if (len == 0) out.append("\"\"")
     else if (s.charAt(len - 1) == '\n') {
-      val splits = YamlRenderer.newlinePattern.split(s)
+      val splits = Platform.getPatternFromCache("\n").split(s.toString)
       out.append('|')
       depth += 1
       splits.foreach { split =>

sjsonnet/test/src/sjsonnet/StdRegexTests.scala

@@ -0,0 +1,38 @@
+package sjsonnet
+
+import sjsonnet.TestUtils.eval
+import utest._
+
+object StdRegexTests extends TestSuite {
+  def tests: Tests = Tests {
+    test("std.native - regex") {
+      eval("""std.native("regexPartialMatch")("a(b)c", "cabc")""") ==> ujson.Obj(
+        "string" -> "abc",
+        "captures" -> ujson.Arr("b")
+      )
+      eval("""std.native("regexPartialMatch")("a(b)c", "def")""") ==> ujson.Obj(
+        "string" -> ujson.Null,
+        "captures" -> ujson.Arr()
+      )
+      eval("""std.native("regexPartialMatch")("a(b)c", "abcabc")""") ==> ujson.Obj(
+        "string" -> "abc",
+        "captures" -> ujson.Arr("b", "b")
+      )
+      eval("""std.native("regexFullMatch")("a(b)c", "abc")""") ==> ujson.Obj(
+        "string" -> "abc",
+        "captures" -> ujson.Arr("b")
+      )
+      eval("""std.native("regexFullMatch")("a(b)c", "cabc")""") ==> ujson.Obj(
+        "string" -> ujson.Null,
+        "captures" -> ujson.Arr()
+      )
+      eval("""std.native("regexFullMatch")("a(b)c", "def")""") ==> ujson.Obj(
+        "string" -> ujson.Null,
+        "captures" -> ujson.Arr()
+      )
+      eval("""std.native("regexGlobalReplace")("abcbbb", "b", "d")""") ==> ujson.Str("adcddd")
+      eval("""std.native("regexReplace")("abcbbb", "b", "d")""") ==> ujson.Str("adcbbb")
+      eval("""std.native("regexQuoteMeta")("a.b")""") ==> ujson.Str(Platform.regexQuote("a.b"))
+    }
+  }
+}