Github user marmbrus commented on a diff in the pull request:
https://github.com/apache/spark/pull/1586#discussion_r15548348
--- Diff:
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/stringOperations.scala
---
@@ -209,6 +212,82 @@ case class EndsWith(left: Expression, right:
Expression)
}
/**
+ * A function that returns the number of bytes in an expression
+ */
+case class Length(child: Expression) extends UnaryExpression {
+
+ type EvaluatedType = Any
+
+ override def dataType = IntegerType
+
+ override def foldable = child.foldable
+
+ override def nullable = child.nullable
+
+ override def toString = s"Length($child)"
+
+ override def eval(input: Row): EvaluatedType = {
+ val string = child.eval(input)
+ if (string == null) {
+ null
+ } else if (!string.isInstanceOf[String]) {
+ string.toString.length
+ } else {
+ new String(string.toString.getBytes,
StrlenConstants.DefaultEncoding).length
+ }
+ }
+
+}
+
+object StrlenConstants {
+ val DefaultEncoding = "ISO-8859-1"
+}
+
+/**
+ * A function that returns the number of characters in a string expression
+ */
+case class Strlen(child: Expression, encoding : Expression) extends
UnaryExpression
+ with Logging {
+
+ type EvaluatedType = Any
+
+ override def dataType = IntegerType
+
+ override def foldable = child.foldable
+
+ override def nullable = true
+
+ override def toString = s"Strlen($child, $encoding)"
+
+ override def eval(input: Row): EvaluatedType = {
+ val string = child.eval(input)
+ if (string == null) {
+ null
+ } else if (!string.isInstanceOf[String]) {
+ log.debug(s"Non-string value [$string] provided to strlen")
+ null
+ } else {
+ var evalEncoding = encoding.eval(input)
+ val strEncoding =
+ if (evalEncoding != null) {
+ evalEncoding.toString
+ } else {
+ StrlenConstants.DefaultEncoding
+ }
+ val s: String = ""
+ try {
+ new String(string.asInstanceOf[String].getBytes,
strEncoding).length
+ } catch {
+ case ue : UnsupportedEncodingException => {
+ log.debug(s"strlen: Caught UnsupportedEncodingException for
encoding=[$strEncoding]")
--- End diff --
This is tricky because it could go either way. In general I have been
deciding based on the following criteria:
- When the input will usually be something static, where an invalid value
is likely a mistake in a query that breaks it for all input, then we throw an
exception (for example invalid regexes).
- When the input is data dependent and is probably something wrong with
the data, return null (for example a divide by zero error, or a failed cast to
a numeric type).
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
