Updata upstream

R/pkg/R/functions.R

@@ -2438,12 +2438,12 @@ setMethod("date_format", signature(y = "Column", x = "character"),
 #' from_json
 #'
 #' Parses a column containing a JSON string into a Column of \code{structType} with the specified
-#' \code{schema} or array of \code{structType} if \code{asJsonArray} is set to \code{TRUE}.
+#' \code{schema} or array of \code{structType} if \code{as.json.array} is set to \code{TRUE}.
 #' If the string is unparseable, the Column will contains the value NA.
 #'
 #' @param x Column containing the JSON string.
 #' @param schema a structType object to use as the schema to use when parsing the JSON string.
-#' @param asJsonArray indicating if input string is JSON array of objects or a single object.
+#' @param as.json.array indicating if input string is JSON array of objects or a single object.
 #' @param ... additional named properties to control how the json is parsed, accepts the same
 #'            options as the JSON data source.
 #'
@@ -2459,8 +2459,8 @@ setMethod("date_format", signature(y = "Column", x = "character"),
 #'}
 #' @note from_json since 2.2.0
 setMethod("from_json", signature(x = "Column", schema = "structType"),
-          function(x, schema, asJsonArray = FALSE, ...) {
-            if (asJsonArray) {
+          function(x, schema, as.json.array = FALSE, ...) {
+            if (as.json.array) {
               jschema <- callJStatic("org.apache.spark.sql.types.DataTypes",
                                      "createArrayType",
                                      schema$jobj)

R/pkg/inst/tests/testthat/test_sparkSQL.R

@@ -1454,7 +1454,7 @@ test_that("column functions", {
   jsonArr <- "[{\"name\":\"Bob\"}, {\"name\":\"Alice\"}]"
   df <- as.DataFrame(list(list("people" = jsonArr)))
   schema <- structType(structField("name", "string"))
-  arr <- collect(select(df, alias(from_json(df$people, schema, asJsonArray = TRUE), "arrcol")))
+  arr <- collect(select(df, alias(from_json(df$people, schema, as.json.array = TRUE), "arrcol")))
   expect_equal(ncol(arr), 1)
   expect_equal(nrow(arr), 1)
   expect_is(arr[[1]][[1]], "list")

resource-managers/yarn/src/main/scala/org/apache/spark/deploy/yarn/Client.scala

@@ -577,7 +577,7 @@ private[spark] class Client(
     ).foreach { case (flist, resType, addToClasspath) =>
       flist.foreach { file =>
         val (_, localizedPath) = distribute(file, resType = resType)
-        // If addToClassPath, we ignore adding jar multiple times to distitrbuted cache.
+        // If addToClassPath, we ignore adding jar multiple times to distributed cache.
         if (addToClasspath) {
           if (localizedPath != null) {
             cachedSecondaryJarLinks += localizedPath

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/catalog/SessionCatalog.scala

@@ -22,6 +22,7 @@ import java.util.Locale
 import javax.annotation.concurrent.GuardedBy
 
 import scala.collection.mutable
+import scala.util.{Failure, Success, Try}
 
 import com.google.common.cache.{Cache, CacheBuilder}
 import org.apache.hadoop.conf.Configuration
@@ -1202,15 +1203,25 @@ class SessionCatalog(
   def listFunctions(db: String, pattern: String): Seq[(FunctionIdentifier, String)] = {
     val dbName = formatDatabaseName(db)
     requireDbExists(dbName)
-    val dbFunctions = externalCatalog.listFunctions(dbName, pattern)
-      .map { f => FunctionIdentifier(f, Some(dbName)) }
-    val loadedFunctions = StringUtils.filterPattern(functionRegistry.listFunction(), pattern)
-      .map { f => FunctionIdentifier(f) }
+    val dbFunctions = externalCatalog.listFunctions(dbName, pattern).map { f =>
+      FunctionIdentifier(f, Some(dbName)) }
+    val loadedFunctions =
+      StringUtils.filterPattern(functionRegistry.listFunction(), pattern).map { f =>
+        // In functionRegistry, function names are stored as an unquoted format.
+        Try(parser.parseFunctionIdentifier(f)) match {
+          case Success(e) => e
+          case Failure(_) =>
+            // The names of some built-in functions are not parsable by our parser, e.g., %
+            FunctionIdentifier(f)
+        }
+      }
     val functions = dbFunctions ++ loadedFunctions
+    // The session catalog caches some persistent functions in the FunctionRegistry
+    // so there can be duplicates.
     functions.map {
       case f if FunctionRegistry.functionSet.contains(f.funcName) => (f, "SYSTEM")
       case f => (f, "USER")
-    }
+    }.distinct
   }
 
 

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/regexpExpressions.scala

@@ -69,7 +69,30 @@ abstract class StringRegexExpression extends BinaryExpression
  * Simple RegEx pattern matching function
  */
 @ExpressionDescription(
-  usage = "str _FUNC_ pattern - Returns true if `str` matches `pattern`, or false otherwise.")
+  usage = "str _FUNC_ pattern - Returns true if str matches pattern, " +
+    "null if any arguments are null, false otherwise.",
+  extended = """
+    Arguments:
+      str - a string expression
+      pattern - a string expression. The pattern is a string which is matched literally, with
+        exception to the following special symbols:
+
+          _ matches any one character in the input (similar to . in posix regular expressions)
+
+          % matches zero ore more characters in the input (similar to .* in posix regular
+          expressions)
+
+        The escape character is '\'. If an escape character precedes a special symbol or another
+        escape character, the following character is matched literally. It is invalid to escape
+        any other character.
+
+    Examples:
+      > SELECT '%SystemDrive%\Users\John' _FUNC_ '\%SystemDrive\%\\Users%'
+      true
+
+    See also:
+      Use RLIKE to match with standard regular expressions.
+""")
 case class Like(left: Expression, right: Expression) extends StringRegexExpression {
 
   override def escape(v: String): String = StringUtils.escapeLikeRegex(v)

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/util/StringUtils.scala

@@ -19,32 +19,44 @@ package org.apache.spark.sql.catalyst.util
 
 import java.util.regex.{Pattern, PatternSyntaxException}
 
+import org.apache.spark.sql.AnalysisException
 import org.apache.spark.unsafe.types.UTF8String
 
 object StringUtils {
 
-  // replace the _ with .{1} exactly match 1 time of any character
-  // replace the % with .*, match 0 or more times with any character
-  def escapeLikeRegex(v: String): String = {
-    if (!v.isEmpty) {
-      "(?s)" + (' ' +: v.init).zip(v).flatMap {
-        case (prev, '\\') => ""
-        case ('\\', c) =>
-          c match {
-            case '_' => "_"
-            case '%' => "%"
-            case _ => Pattern.quote("\\" + c)
-          }
-        case (prev, c) =>
+  /**
+   * Validate and convert SQL 'like' pattern to a Java regular expression.
+   *
+   * Underscores (_) are converted to '.' and percent signs (%) are converted to '.*', other
+   * characters are quoted literally. Escaping is done according to the rules specified in
+   * [[org.apache.spark.sql.catalyst.expressions.Like]] usage documentation. An invalid pattern will
+   * throw an [[AnalysisException]].
+   *
+   * @param pattern the SQL pattern to convert
+   * @return the equivalent Java regular expression of the pattern
+   */
+  def escapeLikeRegex(pattern: String): String = {
+    val in = pattern.toIterator
+    val out = new StringBuilder()
+
+    def fail(message: String) = throw new AnalysisException(
+      s"the pattern '$pattern' is invalid, $message")
+
+    while (in.hasNext) {
+      in.next match {
+        case '\\' if in.hasNext =>
+          val c = in.next
           c match {
-            case '_' => "."
-            case '%' => ".*"
-            case _ => Pattern.quote(Character.toString(c))
+            case '_' | '%' | '\\' => out ++= Pattern.quote(Character.toString(c))
+            case _ => fail(s"the escape character is not allowed to precede '$c'")
           }
-      }.mkString
-    } else {
-      v
+        case '\\' => fail("it is not allowed to end with the escape character")
+        case '_' => out ++= "."
+        case '%' => out ++= ".*"
+        case c => out ++= Pattern.quote(Character.toString(c))
+      }
     }
+    "(?s)" + out.result() // (?s) enables dotall mode, causing "." to match new lines
   }
 
   private[this] val trueStrings = Set("t", "true", "y", "yes", "1").map(UTF8String.fromString)

sql/catalyst/src/test/scala/org/apache/spark/sql/catalyst/expressions/RegexpExpressionsSuite.scala

@@ -18,16 +18,38 @@
 package org.apache.spark.sql.catalyst.expressions
 
 import org.apache.spark.SparkFunSuite
+import org.apache.spark.sql.AnalysisException
 import org.apache.spark.sql.catalyst.dsl.expressions._
-import org.apache.spark.sql.types.StringType
+import org.apache.spark.sql.types.{IntegerType, StringType}
 
 /**
  * Unit tests for regular expression (regexp) related SQL expressions.
  */
 class RegexpExpressionsSuite extends SparkFunSuite with ExpressionEvalHelper {
 
-  test("LIKE literal Regular Expression") {
-    checkEvaluation(Literal.create(null, StringType).like("a"), null)
+  /**
+   * Check if a given expression evaluates to an expected output, in case the input is
+   * a literal and in case the input is in the form of a row.
+   * @tparam A type of input
+   * @param mkExpr the expression to test for a given input
+   * @param input value that will be used to create the expression, as literal and in the form
+   *        of a row
+   * @param expected the expected output of the expression
+   * @param inputToExpression an implicit conversion from the input type to its corresponding
+   *        sql expression
+   */
+  def checkLiteralRow[A](mkExpr: Expression => Expression, input: A, expected: Any)
+    (implicit inputToExpression: A => Expression): Unit = {
+    checkEvaluation(mkExpr(input), expected) // check literal input
+
+    val regex = 'a.string.at(0)
+    checkEvaluation(mkExpr(regex), expected, create_row(input)) // check row input
+  }
+
+  test("LIKE Pattern") {
+
+    // null handling
+    checkLiteralRow(Literal.create(null, StringType).like(_), "a", null)
     checkEvaluation(Literal.create("a", StringType).like(Literal.create(null, StringType)), null)
     checkEvaluation(Literal.create(null, StringType).like(Literal.create(null, StringType)), null)
     checkEvaluation(
@@ -39,45 +61,64 @@ class RegexpExpressionsSuite extends SparkFunSuite with ExpressionEvalHelper {
     checkEvaluation(
       Literal.create(null, StringType).like(NonFoldableLiteral.create(null, StringType)), null)
 
-    checkEvaluation("abdef" like "abdef", true)
-    checkEvaluation("a_%b" like "a\\__b", true)
-    checkEvaluation("addb" like "a_%b", true)
-    checkEvaluation("addb" like "a\\__b", false)
-    checkEvaluation("addb" like "a%\\%b", false)
-    checkEvaluation("a_%b" like "a%\\%b", true)
-    checkEvaluation("addb" like "a%", true)
-    checkEvaluation("addb" like "**", false)
-    checkEvaluation("abc" like "a%", true)
-    checkEvaluation("abc"  like "b%", false)
-    checkEvaluation("abc"  like "bc%", false)
-    checkEvaluation("a\nb" like "a_b", true)
-    checkEvaluation("ab" like "a%b", true)
-    checkEvaluation("a\nb" like "a%b", true)
-  }
+    // simple patterns
+    checkLiteralRow("abdef" like _, "abdef", true)
+    checkLiteralRow("a_%b" like _, "a\\__b", true)
+    checkLiteralRow("addb" like _, "a_%b", true)
+    checkLiteralRow("addb" like _, "a\\__b", false)
+    checkLiteralRow("addb" like _, "a%\\%b", false)
+    checkLiteralRow("a_%b" like _, "a%\\%b", true)
+    checkLiteralRow("addb" like _, "a%", true)
+    checkLiteralRow("addb" like _, "**", false)
+    checkLiteralRow("abc" like _, "a%", true)
+    checkLiteralRow("abc"  like _, "b%", false)
+    checkLiteralRow("abc"  like _, "bc%", false)
+    checkLiteralRow("a\nb" like _, "a_b", true)
+    checkLiteralRow("ab" like _, "a%b", true)
+    checkLiteralRow("a\nb" like _, "a%b", true)
+
+    // empty input
+    checkLiteralRow("" like _, "", true)
+    checkLiteralRow("a" like _, "", false)
+    checkLiteralRow("" like _, "a", false)
+
+    // SI-17647 double-escaping backslash
+    checkLiteralRow("""\\\\""" like _, """%\\%""", true)
+    checkLiteralRow("""%%""" like _, """%%""", true)
+    checkLiteralRow("""\__""" like _, """\\\__""", true)
+    checkLiteralRow("""\\\__""" like _, """%\\%\%""", false)
+    checkLiteralRow("""_\\\%""" like _, """%\\""", false)
+
+    // unicode
+    // scalastyle:off nonascii
+    checkLiteralRow("a\u20ACa" like _, "_\u20AC_", true)
+    checkLiteralRow("a€a" like _, "_€_", true)
+    checkLiteralRow("a€a" like _, "_\u20AC_", true)
+    checkLiteralRow("a\u20ACa" like _, "_€_", true)
+    // scalastyle:on nonascii
+
+    // invalid escaping
+    val invalidEscape = intercept[AnalysisException] {
+      evaluate("""a""" like """\a""")
+    }
+    assert(invalidEscape.getMessage.contains("pattern"))
+
+    val endEscape = intercept[AnalysisException] {
+      evaluate("""a""" like """a\""")
+    }
+    assert(endEscape.getMessage.contains("pattern"))
+
+    // case
+    checkLiteralRow("A" like _, "a%", false)
+    checkLiteralRow("a" like _, "A%", false)
+    checkLiteralRow("AaA" like _, "_a_", true)
 
-  test("LIKE Non-literal Regular Expression") {
-    val regEx = 'a.string.at(0)
-    checkEvaluation("abcd" like regEx, null, create_row(null))
-    checkEvaluation("abdef" like regEx, true, create_row("abdef"))
-    checkEvaluation("a_%b" like regEx, true, create_row("a\\__b"))
-    checkEvaluation("addb" like regEx, true, create_row("a_%b"))
-    checkEvaluation("addb" like regEx, false, create_row("a\\__b"))
-    checkEvaluation("addb" like regEx, false, create_row("a%\\%b"))
-    checkEvaluation("a_%b" like regEx, true, create_row("a%\\%b"))
-    checkEvaluation("addb" like regEx, true, create_row("a%"))
-    checkEvaluation("addb" like regEx, false, create_row("**"))
-    checkEvaluation("abc" like regEx, true, create_row("a%"))
-    checkEvaluation("abc" like regEx, false, create_row("b%"))
-    checkEvaluation("abc" like regEx, false, create_row("bc%"))
-    checkEvaluation("a\nb" like regEx, true, create_row("a_b"))
-    checkEvaluation("ab" like regEx, true, create_row("a%b"))
-    checkEvaluation("a\nb" like regEx, true, create_row("a%b"))
-
-    checkEvaluation(Literal.create(null, StringType) like regEx, null, create_row("bc%"))
+    // example
+    checkLiteralRow("""%SystemDrive%\Users\John""" like _, """\%SystemDrive\%\\Users%""", true)
   }
 
-  test("RLIKE literal Regular Expression") {
-    checkEvaluation(Literal.create(null, StringType) rlike "abdef", null)
+  test("RLIKE Regular Expression") {
+    checkLiteralRow(Literal.create(null, StringType) rlike _, "abdef", null)
     checkEvaluation("abdef" rlike Literal.create(null, StringType), null)
     checkEvaluation(Literal.create(null, StringType) rlike Literal.create(null, StringType), null)
     checkEvaluation("abdef" rlike NonFoldableLiteral.create("abdef", StringType), true)
@@ -87,42 +128,32 @@ class RegexpExpressionsSuite extends SparkFunSuite with ExpressionEvalHelper {
     checkEvaluation(
       Literal.create(null, StringType) rlike NonFoldableLiteral.create(null, StringType), null)
 
-    checkEvaluation("abdef" rlike "abdef", true)
-    checkEvaluation("abbbbc" rlike "a.*c", true)
+    checkLiteralRow("abdef" rlike _, "abdef", true)
+    checkLiteralRow("abbbbc" rlike _, "a.*c", true)
 
-    checkEvaluation("fofo" rlike "^fo", true)
-    checkEvaluation("fo\no" rlike "^fo\no$", true)
-    checkEvaluation("Bn" rlike "^Ba*n", true)
-    checkEvaluation("afofo" rlike "fo", true)
-    checkEvaluation("afofo" rlike "^fo", false)
-    checkEvaluation("Baan" rlike "^Ba?n", false)
-    checkEvaluation("axe" rlike "pi|apa", false)
-    checkEvaluation("pip" rlike "^(pi)*$", false)
+    checkLiteralRow("fofo" rlike _, "^fo", true)
+    checkLiteralRow("fo\no" rlike _, "^fo\no$", true)
+    checkLiteralRow("Bn" rlike _, "^Ba*n", true)
+    checkLiteralRow("afofo" rlike _, "fo", true)
+    checkLiteralRow("afofo" rlike _, "^fo", false)
+    checkLiteralRow("Baan" rlike _, "^Ba?n", false)
+    checkLiteralRow("axe" rlike _, "pi|apa", false)
+    checkLiteralRow("pip" rlike _, "^(pi)*$", false)
 
-    checkEvaluation("abc"  rlike "^ab", true)
-    checkEvaluation("abc"  rlike "^bc", false)
-    checkEvaluation("abc"  rlike "^ab", true)
-    checkEvaluation("abc"  rlike "^bc", false)
+    checkLiteralRow("abc"  rlike _, "^ab", true)
+    checkLiteralRow("abc"  rlike _, "^bc", false)
+    checkLiteralRow("abc"  rlike _, "^ab", true)
+    checkLiteralRow("abc"  rlike _, "^bc", false)
 
     intercept[java.util.regex.PatternSyntaxException] {
       evaluate("abbbbc" rlike "**")
     }
-  }
-
-  test("RLIKE Non-literal Regular Expression") {
-    val regEx = 'a.string.at(0)
-    checkEvaluation("abdef" rlike regEx, true, create_row("abdef"))
-    checkEvaluation("abbbbc" rlike regEx, true, create_row("a.*c"))
-    checkEvaluation("fofo" rlike regEx, true, create_row("^fo"))
-    checkEvaluation("fo\no" rlike regEx, true, create_row("^fo\no$"))
-    checkEvaluation("Bn" rlike regEx, true, create_row("^Ba*n"))
-
     intercept[java.util.regex.PatternSyntaxException] {
-      evaluate("abbbbc" rlike regEx, create_row("**"))
+      val regex = 'a.string.at(0)
+      evaluate("abbbbc" rlike regex, create_row("**"))
     }
   }
 
-
   test("RegexReplace") {
     val row1 = create_row("100-200", "(\\d+)", "num")
     val row2 = create_row("100-200", "(\\d+)", "###")