languagetool-org/languagetool · error

Expected semicolon-separated input:

Error message

Expected semicolon-separated input: 

What it means

AutomaticProhibitedCompoundRuleEvaluator.run() reads input lines that must contain exactly two semicolon-separated fields (a compound pair). Lines with comments (#) are skipped, but any remaining line whose split on ";\s*" does not yield exactly 2 parts triggers this IOException, because pairwise comparison of the compound variants is impossible.

Source

Thrown at languagetool-dev/src/main/java/org/languagetool/dev/bigdata/AutomaticProhibitedCompoundRuleEvaluator.java:82

    DirectoryReader reader = DirectoryReader.open(FSDirectory.open(luceneIndexDir.toPath()));
    searcher = new IndexSearcher(reader);
    InputStream confusionSetStream = JLanguageTool.getDataBroker().getFromResourceDirAsStream("/" + LANGUAGE + "/confusion_sets.txt");
    knownSets = new ConfusionSetLoader(language).loadConfusionPairs(confusionSetStream);
  }

  private void run(List<String> lines, File indexDir) throws IOException {
    LanguageModel lm = new LuceneLanguageModel(indexDir);
    ProhibitedCompoundRuleEvaluator evaluator = new ProhibitedCompoundRuleEvaluator(language, lm);
    int lineCount = 0;
    for (String line : lines) {
      lineCount++;
      if (line.contains("#")) {
        System.out.println("Ignoring: " + line);
        continue;
      }
      String[] parts = line.split(";\\s*");
      if (parts.length != 2) {
        throw new IOException("Expected semicolon-separated input: " + line);
      }
      try {
        int i = 1;
        for (String part : parts) {
          // compare pair-wise - maybe we should compare every item with every other item?
          if (i < parts.length) {
            runOnPair(evaluator, line, lineCount, lines.size(), removeComment(part), removeComment(parts[i]));
          }
          i++;
        }
      } catch (RuntimeException e) {
        e.printStackTrace();
      }
    }
    System.out.println("Done. Ignored items because they are already known: " + ignored);
  }

  private String removeComment(String str) {

View on GitHub (pinned to 2e990059ce)

Solutions

1. Inspect the offending line printed in the message and correct it to exactly two semicolon-separated fields: word1;word2.
2. Trim the line and skip blank lines before splitting.
3. Handle comment lines robustly (skip lines starting with # or strip trailing comments) so partial comments don't corrupt the field count.
4. Make the parser tolerant: allow >=2 parts, or log-and-continue on malformed lines.

Example fix

// before
String[] parts = line.split(";\\s*");
if (parts.length != 2) {
  throw new IOException("Expected semicolon-separated input: " + line);
}
// after
String trimmed = line.trim();
if (trimmed.isEmpty() || trimmed.startsWith("#")) continue;
String[] parts = trimmed.split(";\\s*");
if (parts.length != 2) {
  System.err.println("WARN: skipping malformed line: " + line);
  continue;
}

String trimmed = line == null ? "" : line.trim();
boolean valid = !trimmed.isEmpty() && !trimmed.startsWith("#")
    && trimmed.split(";\\s*").length == 2;
if (!valid) throw new IllegalArgumentException("Bad compound pair line: " + line);

try {
  evaluator.run(inputPath);
} catch (IOException e) {
  if (e.getMessage().startsWith("Expected semicolon-separated input")) {
    System.err.println("Fix line format (word1;word2): " + e.getMessage());
  }
}

When it happens

Trigger: An input line that isn't of the form 'word1;word2' — e.g. a single value with no semicolon, three or more semicolon-separated values, an empty or whitespace-only line slipping through, or a line whose '#' comment appears not at start so it isn't filtered by the contains("#") check but changes field count expectations.

Common situations: Hand-edited compound lists with stray semicolons; CSV exported with commas instead of semicolons; trailing semicolons producing an empty third part; pasting lines with embedded comments or extra columns.

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

• Could not load simple replacement data from: " + path + ". E
• Format error in file " + path + ", line: " + line
• Cannot load or parse input stream of '${filename}'
• No ngram data found for:
• Error: Lines from the input file should contain at least two

AI-assisted analysis of languagetool-org/languagetool@2e990059ce (2026-09-06). Data as JSON: /api/errors/31853dfcfe189046. Report an issue: GitHub.