NationalSecurityAgency/ghidra · error · IOException

Bad line in score file

Error message

Bad line in score file

What it means

Thrown by FileScoreCaching.loadCache() when a score entry line (after the two threshold headers) does not conform to the expected format: exactly two space-separated tokens where the first token is a 32-character md5 string and the second is a float. Any line failing split.length != 2 or split[0].length() != 32 triggers this.

Source

Thrown at Ghidra/Features/BSim/src/main/java/ghidra/features/bsim/query/client/FileScoreCaching.java:64

		BufferedReader reader = new BufferedReader(new FileReader(storageFile));
		String floatString = reader.readLine();
		if (floatString == null) {
			reader.close();
			throw new IOException("Score file missing threshold lines");
		}
		simThreshold = Float.parseFloat(floatString);
		floatString = reader.readLine();
		if (floatString == null) {
			reader.close();
			throw new IOException("Score file missing threshold lines");
		}
		sigThreshold = Float.parseFloat(floatString);
		floatString = reader.readLine();
		while (floatString != null) {
			String[] split = floatString.split(" ");
			if (split.length != 2 || split[0].length() != 32) {
				reader.close();
				throw new IOException("Bad line in score file");
			}
			float val = Float.parseFloat(split[1]);
			cacheMap.put(split[0], val);
			floatString = reader.readLine();
		}
		reader.close();
	}

	@Override
	public float getSelfScore(String md5) throws LSHException {
		try {
			loadCache();
		}
		catch (IOException e) {
			throw new LSHException("Could not recover cached scores: " + e.getMessage());
		}
		if (cacheMap != null) {
			Float val = cacheMap.get(md5);

View on GitHub (pinned to d5f144c24d)

Solutions

1. Delete the cache file and regenerate it by running the scoring workflow that commits self-scores.
2. Call resetStorage(simThresh, sigThresh) to wipe and reinitialize.
3. If you must repair, verify each data line is "<32-char-md5> <float>" with a single space separator and LF line endings.

Example fix

// before — cache file has a bad line: "abc123 0.75" (md5 too short)
FileScoreCaching cache = new FileScoreCaching(path);
cache.getSelfScore(md5); // throws "Bad line in score file"

// after
new File(path).delete(); // or cache.resetStorage(simThresh, sigThresh)
// re-run the ingest that populates self-scores

File cacheFile = new File(cachePath);
if (cacheFile.exists()) {
    try (BufferedReader r = new BufferedReader(new FileReader(cacheFile))) {
        r.readLine(); r.readLine(); // skip thresholds
        String line;
        while ((line = r.readLine()) != null) {
            String[] parts = line.split(" ");
            if (parts.length != 2 || parts[0].length() != 32) {
                cacheFile.delete(); // malformed
                break;
            }
        }
    }
}

try {
    cache.getSelfScore(md5);
} catch (LSHException e) {
    if (e.getMessage().contains("Bad line")) {
        cache.resetStorage(simThresh, sigThresh);
    } else throw e;
}

When it happens

Trigger: Loading a cache file where a data line is malformed — e.g. a tab instead of a space delimiter, a truncated md5, extra whitespace producing more than 2 split tokens, a non-numeric score value, or a blank line in the middle of the data section.

Common situations: The cache file was hand-edited or generated by a different tool. A filesystem encoding issue or line-ending mismatch (CRLF vs LF) corrupted the split. An older BSim version wrote a different format that is now incompatible.

Related errors

• Score file missing threshold lines
• Could not recover cached scores: {e.getMessage()}
• Self-score not recorded for {md5}
• Could not commit self-score: {ex.getMessage()}
• Problems loading score cache: {e.getMessage()}

AI-assisted analysis of NationalSecurityAgency/ghidra@d5f144c24d (2026-08-14). Data as JSON: /api/errors/19cce746311942f0. Report an issue: GitHub.