[SPARK-18952] regex strings not properly escaped in codegen for aggregations

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.0.2
• Fix Version/s: 2.1.1, 2.2.0
• Component/s: SQL
• Assignee: Burak Yavuz
• Reporter: Burak Yavuz
• Created: 20/Dec/16 21:59
• Updated: 02/May/17 18:30
• Resolved: 09/Jan/17 23:18

Description

If I use the function regexp_extract, and then in my regex string, use `\`, i.e. escape character, this fails codegen, because the `\` character is not properly escaped when codegen'd.

Example stack trace:

/* 059 */     private int maxSteps = 2;
/* 060 */     private int numRows = 0;
/* 061 */     private org.apache.spark.sql.types.StructType keySchema = new org.apache.spark.sql.types.StructType().add("date_format(window#325.start, yyyy-MM-dd HH:mm)", org.apache.spark.sql.types.DataTypes.StringType)
/* 062 */     .add("regexp_extract(source#310.description, ([a-zA-Z]+)\[.*, 1)", org.apache.spark.sql.types.DataTypes.StringType);
/* 063 */     private org.apache.spark.sql.types.StructType valueSchema = new org.apache.spark.sql.types.StructType().add("sum", org.apache.spark.sql.types.DataTypes.LongType);
/* 064 */     private Object emptyVBase;

...

org.codehaus.commons.compiler.CompileException: File 'generated.java', Line 62, Column 58: Invalid escape sequence
	at org.codehaus.janino.Scanner.scanLiteralCharacter(Scanner.java:918)
	at org.codehaus.janino.Scanner.produce(Scanner.java:604)
	at org.codehaus.janino.Parser.peekRead(Parser.java:3239)
	at org.codehaus.janino.Parser.parseArguments(Parser.java:3055)
	at org.codehaus.janino.Parser.parseSelector(Parser.java:2914)
	at org.codehaus.janino.Parser.parseUnaryExpression(Parser.java:2617)
	at org.codehaus.janino.Parser.parseMultiplicativeExpression(Parser.java:2573)
	at org.codehaus.janino.Parser.parseAdditiveExpression(Parser.java:2552)

In the codegend expression, the literal should use `
` instead of `\`

Issue Links

• duplicates: SPARK-18866 Codegen fails with cryptic error if regexp_replace() output column is not aliased
• is duplicated by: SPARK-20556 codehaus fails to generate code because of unescaped strings
• links to: [Github] Pull Request #16361 (brkyvz)
• links to: [Github] Pull Request #16518 (brkyvz)