[SPARK-12451] Regexp functions don't support patterns containing '*/'

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 1.5.2
• Component/s: SQL
• Assignee: Unassigned
• Reporter: William Dee
• Created: 21/Dec/15 16:58
• Updated: 24/Oct/16 04:47
• Resolved: 24/Oct/16 04:47

Description

When using the regexp functions in Spark SQL, patterns containing '*/' create runtime errors in the auto generated code. This is due to the fact that the code generator creates a multiline comment containing, amongst other things, the pattern.

Here is an excerpt from my stacktrace to illustrate: (Helpfully, the stack trace includes all of the auto-generated code)

Caused by: org.codehaus.commons.compiler.CompileException: Line 232, Column 54: Unexpected token "," in primary
	at org.codehaus.janino.Parser.compileException(Parser.java:3125)
	at org.codehaus.janino.Parser.parsePrimary(Parser.java:2512)
	at org.codehaus.janino.Parser.parseUnaryExpression(Parser.java:2252)
	at org.codehaus.janino.Parser.parseMultiplicativeExpression(Parser.java:2211)
	at org.codehaus.janino.Parser.parseAdditiveExpression(Parser.java:2190)
	at org.codehaus.janino.Parser.parseShiftExpression(Parser.java:2169)
	at org.codehaus.janino.Parser.parseRelationalExpression(Parser.java:2072)
	at org.codehaus.janino.Parser.parseEqualityExpression(Parser.java:2046)
	at org.codehaus.janino.Parser.parseAndExpression(Parser.java:2025)
	at org.codehaus.janino.Parser.parseExclusiveOrExpression(Parser.java:2004)
	at org.codehaus.janino.Parser.parseInclusiveOrExpression(Parser.java:1983)
	at org.codehaus.janino.Parser.parseConditionalAndExpression(Parser.java:1962)
	at org.codehaus.janino.Parser.parseConditionalOrExpression(Parser.java:1941)
	at org.codehaus.janino.Parser.parseConditionalExpression(Parser.java:1922)
	at org.codehaus.janino.Parser.parseAssignmentExpression(Parser.java:1901)
	at org.codehaus.janino.Parser.parseExpression(Parser.java:1886)
	at org.codehaus.janino.Parser.parseBlockStatement(Parser.java:1149)
	at org.codehaus.janino.Parser.parseBlockStatements(Parser.java:1085)
	at org.codehaus.janino.Parser.parseMethodDeclarationRest(Parser.java:938)
	at org.codehaus.janino.Parser.parseClassBodyDeclaration(Parser.java:620)
	at org.codehaus.janino.Parser.parseClassBody(Parser.java:515)
	at org.codehaus.janino.Parser.parseClassDeclarationRest(Parser.java:481)
	at org.codehaus.janino.Parser.parseClassBodyDeclaration(Parser.java:577)
	at org.codehaus.janino.ClassBodyEvaluator.cook(ClassBodyEvaluator.java:229)
	at org.codehaus.janino.SimpleCompiler.cook(SimpleCompiler.java:192)
	at org.codehaus.commons.compiler.Cookable.cook(Cookable.java:84)
	at org.codehaus.commons.compiler.Cookable.cook(Cookable.java:77)
	at org.apache.spark.sql.catalyst.expressions.codegen.CodeGenerator.org$apache$spark$sql$catalyst$expressions$codegen$CodeGenerator$$doCompile(CodeGenerator.scala:387)

... line 232 ...

    /* regexp_replace(input[46, StringType],^.*/,) */
    
    /* input[46, StringType] */
    
    boolean isNull31 = i.isNullAt(46);
    UTF8String primitive32 = isNull31 ? null : (i.getUTF8String(46));
    
    boolean isNull24 = true;
    UTF8String primitive25 = null;
    if (!isNull31) {
      /* ^.*/ */
      
      /* expression: ^.*/ */
      Object obj35 = expressions[4].eval(i);
      boolean isNull33 = obj35 == null;
      UTF8String primitive34 = null;
      if (!isNull33) {
        primitive34 = (UTF8String) obj35;
      }
...

Note the multiple multiline comments, these obviously break when the regex pattern contains the end-of-comment token '*/'

Issue Links

• duplicates: SPARK-11352 codegen.GeneratePredicate fails due to unquoted comment