stanfordnlp/CoreNLP · error · RuntimeException

Invalid class in file:

Error message

Invalid class in file:

What it means

When loading a parser from a serialized file, the object stream read failed with java.io.InvalidClassException. LexicalizedParser.getParserFromSerializedFile deliberately wraps this in a RuntimeException ("Invalid class in file: ...") because retrying the file as a text grammar would be pointless. It indicates the serialized class descriptors in the file do not match the classes on the classpath.

Solutions

  1. Align jar and model versions: use a parser model file built with (or known to be compatible with) the exact library version on your classpath
  2. Re-serialize/retrain the parser with the current library version, or re-download the matching pre-trained model from the Stanford model release page
  3. Check for duplicate/stale parser jars (mvn dependency:tree, look in WEB-INF/lib, or jarseverywhere issues) and remove the conflicting one
  4. Print the nested InvalidClassException's serialVersionUID detail to see which class mismatched and restore that class version

Example fix

// before
LexicalizedParser lp = LexicalizedParser.loadModel("model-3.6.0.ser.gz"); // with CoreNLP 4.x jar
// after
LexicalizedParser lp = LexicalizedParser.loadModel("englishPCFG-4.5.0.ser.gz"); // version-matched model
Defensive patterns

Strategy: try-catch

Validate before calling

// Check jar version vs. expected model version before loading
String version = LexicalizedParser.class.getPackage().getImplementationVersion();
// ensure model file was built for this version (versioned filenames help)

Try / catch

try {
  LexicalizedParser lp = LexicalizedParser.loadModel(path);
} catch (RuntimeException e) {
  if (e.getCause() instanceof InvalidClassException) {
    throw new IllegalStateException("Model/library version mismatch for " + path + ": " + e.getCause().getMessage(), e);
  }
  throw e;
}

Prevention

When it happens

Trigger: Calling LexicalizedParser.loadModel/getParserFromSerializedFile on a .ser/.ser.gz model serialized by a different (usually older or newer) version of the parser classes than the ones on the classpath, or on a corrupt file.

Common situations: Upgrading the Stanford Parser/CoreNLP jar while keeping an old trained model; mixing parser jars on the classpath so a stale class shadows the right one; a model trained with a custom TreebankLanguagePack/Options class that was later renamed or changed serial version.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/8b5aff8eff44c20b. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/parser/lexparser/LexicalizedParser.java:577

    } catch (IOException e) {
      e.printStackTrace();
    }
    return null;
  }


  public static LexicalizedParser getParserFromSerializedFile(String serializedFileOrUrl) {
    try {
      Timing tim = new Timing();
      ObjectInputStream in = IOUtils.readStreamFromString(serializedFileOrUrl);
      LexicalizedParser pd = loadModel(in);

      in.close();
      log.info("Loading parser from serialized file " + serializedFileOrUrl + " ... done [" + tim.toSecondsString() + " sec].");
      return pd;
    } catch (InvalidClassException ice) {
      // For this, it's not a good idea to continue and try it as a text file!
      throw new RuntimeException("Invalid class in file: " + serializedFileOrUrl, ice);
    } catch (FileNotFoundException fnfe) {
      // For this, it's not a good idea to continue and try it as a text file!
      throw new RuntimeException("File not found: " + serializedFileOrUrl, fnfe);
    } catch (StreamCorruptedException sce) {
      // suppress error message, on the assumption that we've really got
      // a text grammar, and that'll be tried next
      log.info("Attempting to load " + serializedFileOrUrl +
               " as a serialized grammar caused error below, but this may just be because it's a text grammar!");
      log.info(sce);
    } catch (Exception e) {
      log.error(e);
    }
    return null;
  }


  private static void printOptions(boolean train, Options op) {
    op.display();

View on GitHub (pinned to 1b7edd19c4)