stanfordnlp/CoreNLP · error · RuntimeException

Error on line

Error message

Error on line 

What it means

readDeps wraps each line's parsing in a try block; if anything fails while parsing a dependency line (bad number format, unknown fields, malformed indices) the BufferedReader is closed and the original exception is rethrown wrapped in a RuntimeException prefixed with 'Error on line <N>' so the user can locate the offending input line.

Solutions

1. Inspect the reported line number and fix that line to match the expected 'rel gov child' format.
2. Strip headers/blank lines and verify the delimiter (tab vs. whitespace) matches the reader's expectation.
3. Regenerate the dependency file from the parser, ensuring the run completed successfully.

Example fix

// before (input line)
nsubj the-dog-1 runs-2
// after
nsubj	1	2

try (Stream<String> lines = Files.lines(path)) { IntStream.range(1, (int) lines.count()+1).forEach(i -> assertValidDepLine(lineAt(i))); }

try { scoring = new DependencyScoring(file, ...); } catch (RuntimeException e) { System.err.println(e.getMessage()); /* message names the offending line */ }

When it happens

Trigger: Any malformed line in the dependency input file: missing tabs/columns, non-numeric indices, empty relation fields, or an exception thrown by the grel lookup — encountered while constructing the DependencyScoring.

Common situations: Files with different column conventions (CoNLL vs. simple 3-column format), trailing blank or header lines, truncated output from a crashed parser, Windows line endings or invisible characters breaking the split.

Understand the failure class

Background: "Invalid ... format", "must be in format X", "does not look like a ..." — invalid argument format errors across CLI tools and libraries — this error's family across 17 libraries.

Related errors

• Invalid number of fields, should be >=1 and <= 31
• Found a non-empty line in a tsurgeon section after reading…
• Unknown minimizer
• Unknown clique: " + clique
• Bad number put into wordToNumber. Word is: \"" + input +…

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/7cb63d5039512159. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/trees/DependencyScoring.java:232

        grel = EnglishGrammaticalRelations.getConj(conj);
      }
      if (grel == null) {
        throw new RuntimeException("Unknown grammatical relation '" + depName+"'");
      }

      //Word govWord = new Word(govName.substring(0, govDash));
      IndexedWord govWord = new IndexedWord();
      govWord.setValue(normalizeNumbers(govName));
      govWord.setWord(govWord.value());
      //Word childWord = new Word(childName.substring(0, childDash));
      IndexedWord childWord = new IndexedWord();
      childWord.setValue(normalizeNumbers(childName));
      childWord.setWord(childWord.value());
      TypedDependency dep = new TypedDependencyStringEquality(grel, govWord, childWord);
      deps.add(dep);
      } catch (Exception e) {
        breader.close();
        throw new RuntimeException("Error on line "+breader.getLineNumber()+":\n\n"+e);
      }
    }
    if (deps.size() != 0) {
       readDeps.add(deps);
    }

    //log.info("last: "+readDeps.get(readDeps.size()-1));
    breader.close();
    return readDeps;
  }

  /**
   * Score system typed dependencies
   *
   * @param system
   * @return a triple consisting of (labeled attachment, unlabeled attachment,
   *         label accuracy)
   */

View on GitHub (pinned to 1b7edd19c4)