stanfordnlp/CoreNLP · error · IllegalArgumentException

Cannot output in format

Error message

Cannot output in format ${outputFormat} from the interactive shell

What it means

In StanfordCoreNLPClient's interactive shell, the chosen output format has no implemented printing path for the annotation. Only text/json/xml/html/serialized cases exist; anything else (or SERVLET-style formats) falls to default and throws IllegalArgumentException. The shell cannot honor the requested output format.

Solutions

  1. Set -outputFormat to one the shell supports: text, json, xml, or html
  2. Check the source of StanfordCoreNLPClient (shell loop) for the exact supported case labels in your version
  3. If you need serialized output, write results programmatically via the client API instead of the interactive shell

Example fix

// before
java -cp corenlp.jar edu.stanford.nlp.pipeline.StanfordCoreNLPClient -outputFormat conll -annotators tokenize
// after
java -cp corenlp.jar edu.stanford.nlp.pipeline.StanfordCoreNLPClient -outputFormat json -annotators tokenize
Defensive patterns

Strategy: validation

Validate before calling

Set<String> supported = Set.of("text","json","xml","html");
String fmt = props.getProperty("outputFormat", "json");
if (!supported.contains(fmt)) throw new IllegalArgumentException("Shell supports only: " + supported + ", got: " + fmt);

Try / catch

try {
  client.runShell(props);
} catch (IllegalArgumentException e) {
  if (e.getMessage().contains("from the interactive shell")) {
    props.setProperty("outputFormat", "text");
    client.runShell(props);
  } else throw e;
}

Prevention

When it happens

Trigger: Calling StanfordCoreNLPClient.run() with the interactive shell active while the -outputFormat property is set to a value the shell switch does not handle (e.g. 'json' omitted by build, or an arbitrary string like 'conll').

Common situations: Users passing server-style output formats (e.g. 'serialized', custom formats) to the client shell; typo'd outputFormat values; copy-pasting CLI options between StanfordCoreNLP and StanfordCoreNLPClient.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/aa260284868d4a94. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/pipeline/StanfordCoreNLPClient.java:618

              new XMLOutputter().print(anno, System.out);
              break;
            case JSON:
              new JSONOutputter().print(anno, System.out);
              System.out.println();
              break;
            case CONLL:
              new CoNLLOutputter().print(anno, System.out);
              System.out.println();
              break;
            case TEXT:
              new TextOutputter().print(anno, System.out);
              break;
            case SERIALIZED:
              warn("You probably cannot read the serialized output, so printing in text instead");
              new TextOutputter().print(anno, System.out);
              break;
            default:
              throw new IllegalArgumentException("Cannot output in format " + outputFormat + " from the interactive shell");
          }
        } catch (IOException e) {
          throw new RuntimeIOException(e);
        }
      }
    });
  }

  /**
   * The implementation of what to run on a command-line call of CoreNLPWebClient
   *
   * @throws IOException If any IO problem
   */
  public void run() throws IOException {
    StanfordRedwoodConfiguration.minimalSetup();
    StanfordCoreNLP.OutputFormat outputFormat = StanfordCoreNLP.OutputFormat.valueOf(properties.getProperty("outputFormat", "text").toUpperCase(Locale.ROOT));

    //

View on GitHub (pinned to 1b7edd19c4)