ANTLR errors -- validity of results

[erikmartinessanches]

When running java -jar jplag.jar -l javascript dirs -bc basecode, jplag shows results in browser but also many ANTLR errors, such as

2025-03-27-04:44:52_500 [ERROR] AntlrLoggerErrorListener - ANTLR error - line 23:0 mismatched input '}' expecting {'[', '(', '=', '?', ':', '.', '++', '--', '+', '-', '!', '*', '/', '%', '>>', '<<', '>>>', '<', '>', '<=', '>=', '==', '!=', '===', '!==', '&', '^', '|', '&&', '||', '*=', '/=', '%=', '+=', '-=', '<<=', '>>=', '>>>=', '&=', '^=', '|=', 'instanceof', 'in', 'as', BackTick}

Is the result reliable regardless?

macOS 15.3.2, M1 Pro.

[tsaglam]

It depends. We use different means of parsing files of programming languages. For many languages, we use ANTLR grammars.
When the ANTLR grammar fails to parse a program (e.g., due to a syntax error), it may still return an incorrect parse tree, and JPlag can still (partially) check that program for plagiarism.

However, if a program cannot be parsed at all, this particular program will then be omitted.

For example, consider these two programs:

If I now introduce a syntax error in line 19 (remove bracket, mistype print):

Note that due to the incorrect parsing, some code can no langer be matched. However, it is not the changed line (19), it is a later part of the program. The reason is, that ANTLR returns an incorrect parse tree due to the error.