stanfordnlp/CoreNLP · error · IllegalArgumentException

Not a valid dashes style:

Error message

Not a valid dashes style:

What it means

PTBLexer options parsing rejects an unknown value for the 'dashes' option. The value string is converted to a LexerUtils.DashesEnum constant via valueOf; if it does not name a valid enum constant, an IllegalArgumentException is thrown. This guards against silent misconfiguration of dash tokenization behavior.

Solutions

  1. Use a valid DashesEnum value (e.g. 'none', 'ptb3', 'unicode') in the options string
  2. Check LexerUtils.DashesEnum for the exact accepted constants and their spellings
  3. Trim/case-insensitively normalize the value before passing (valueOf already uppercases, so spelling matters more than case)

Example fix

// before
PTBTokenizer coreLabelTokenizer = new PTBTokenizer<>(reader, "dashes=hyphens", false);
// after
PTBTokenizer coreLabelTokenizer = new PTBTokenizer<>(reader, "dashes=ptb3", false);
Defensive patterns

Strategy: validation

Validate before calling

Set<String> valid = Arrays.stream(LexerUtils.DashesEnum.values()).map(Enum::name).collect(Collectors.toSet());
if (!valid.contains(value.trim().toUpperCase(Locale.ROOT))) throw new IllegalArgumentException("bad dashes option: " + value);

Try / catch

try {
  tokenizer = new PTBTokenizer<>(reader, factory, options);
} catch (IllegalArgumentException e) {
  if (e.getMessage().startsWith("Not a valid dashes style")) { /* fallback to default 'dashes=ptb3' or surface config error */ }
  throw e;
}

Prevention

When it happens

Trigger: Calling PTBLexer (or PTBTokenizer with options) with 'dashes' set to a string that is not a DashesEnum name, e.g. options string 'dashes=hyphens' or 'dashes=emdash' (valid values are typically 'none', 'ptb3', 'unicode').

Common situations: Copy-pasting tokenizer options from old documentation or blog posts; typos in the option value; assuming free-text dash styles are accepted; migrating configs between CoreNLP versions where enum names changed.

Understand the failure class

Background: Invalid enum value errors: "Unknown type", "Invalid scope", "must be one of" — when a string is not on the library's allowed list — this error's family across 23 libraries.

Related errors

AI-assisted analysis of stanfordnlp/CoreNLP@1b7edd19c4 (2026-09-10). Data as JSON: /api/errors/ce1eb097d533452d. Report an issue: GitHub.

Appendix: source

Thrown at src/edu/stanford/nlp/process/PTBLexer.flex:216

              throw new IllegalArgumentException ("Not a valid quotes style: " + value);
            }
          } else if ("splitAssimilations".equals(key)) {
            splitAssimilations = val;
          } else if ("splitHyphenated".equals(key)) {
            splitHyphenated = val;
          } else if ("splitForwardSlash".equals(key)) {
            splitForwardSlash = val;
          } else if ("ellipses".equals(key)) {
            try {
              ellipsisStyle = LexerUtils.EllipsesEnum.valueOf(value.trim().toUpperCase(Locale.ROOT));
            } catch (IllegalArgumentException iae) {
              throw new IllegalArgumentException ("Not a valid ellipses style: " + value);
            }
          } else if ("dashes".equals(key)) {
            try {
              dashesStyle = LexerUtils.DashesEnum.valueOf(value.trim().toUpperCase(Locale.ROOT));
            } catch (IllegalArgumentException iae) {
              throw new IllegalArgumentException ("Not a valid dashes style: " + value);
            }
          } else if ("escapeForwardSlashAsterisk".equals(key)) {
            escapeForwardSlashAsterisk = val;
          } else if ("untokenizable".equals(key)) {
            switch (value) {
              case "noneDelete":
                untokenizable = UntokenizableOptions.NONE_DELETE;
                break;
              case "firstDelete":
                untokenizable = UntokenizableOptions.FIRST_DELETE;
                break;
              case "allDelete":
                untokenizable = UntokenizableOptions.ALL_DELETE;
                break;
              case "noneKeep":
                untokenizable = UntokenizableOptions.NONE_KEEP;
                break;
              case "firstKeep":

View on GitHub (pinned to 1b7edd19c4)