hankcs/HanLP · error · RuntimeError

Conversion failed with code {code} for {src}. The err messag

Error message

Conversion failed with code {code} for {src}. The err message is:\n {err}\nDo you have java installed? Do you have enough memory?

What it means

After running the Stanford Parser jar via subprocess to convert CTB trees, convert_to_dependency checks the process exit code. A non-zero code means java failed (missing JVM, wrong java version, or out-of-memory during conversion), and the collected stderr is included in the RuntimeError.

Source

Thrown at hanlp/datasets/parsing/loaders/_ctb_utils.py:103

    sp_home = get_resource(sp_home)
    # jar_path = get_resource(f'{sp_home}#stanford-parser.jar')
    if ud:
        jclass = 'edu.stanford.nlp.trees.international.pennchinese.UniversalChineseGrammaticalStructure' if language == 'zh' \
            else 'edu.stanford.nlp.trees.ud.UniversalDependenciesConverter'
    else:
        jclass = 'edu.stanford.nlp.trees.international.pennchinese.ChineseGrammaticalStructure' if language == 'zh' \
            else 'edu.stanford.nlp.trees.EnglishGrammaticalStructure'
    cmd = f'java -cp {sp_home}/* {jclass} ' \
          f'-treeFile {src}'
    if conllx:
        cmd += ' -conllx'
    if not ud:
        cmd += f' -basic -keepPunct'
    code, out, err = get_exitcode_stdout_stderr(cmd)
    with open(dst, 'w') as f:
        f.write(out)
    if code:
        raise RuntimeError(f'Conversion failed with code {code} for {src}. The err message is:\n {err}\n'
                           f'Do you have java installed? Do you have enough memory?')


def clean_ctb_bracketed(ctb_root, out_root):
    os.makedirs(out_root, exist_ok=True)
    ctb_root = join(ctb_root, 'bracketed')
    chtbs = _list_treebank_root(ctb_root)
    timer = CountdownTimer(len(chtbs))
    for f in chtbs:
        with open(join(ctb_root, f), encoding='utf-8') as src, open(join(out_root, f + '.txt'), 'w',
                                                                    encoding='utf-8') as out:
            for line in src:
                if not line.strip().startswith('<'):
                    out.write(line)
        timer.log('Cleaning up CTB [blink][yellow]...[/yellow][/blink]', erase=False)


def _list_treebank_root(ctb_root):

View on GitHub (pinned to ddb1299bdd)

Solutions

1. Install a JDK/JRE (java -version must work) — Stanford Parser 4.2.0 needs Java 8+
2. If it's an OOM, raise heap: export JAVA_OPTS or _JAVA_OPTIONS='-Xmx8g' and retry
3. Check the err text in the message for the underlying java error and address it (version mismatch, corrupt download)
4. Verify the Stanford Parser zip downloaded fully (re-download if corrupted)

Example fix

# shell
sudo apt-get install -y openjdk-11-jre-headless
export _JAVA_OPTIONS='-Xmx8g'
python prepare_ctb.py

import shutil
if shutil.which('java') is None:
    raise SystemExit('java not found; install a JDK/JRE before conversion')

try:
    convert_to_dependency(src, dst, language='zh', version='4.2.0')
except RuntimeError as e:
    msg = str(e)
    if 'enough memory' in msg:
        import os
        os.environ['_JAVA_OPTIONS'] = '-Xmx8g'
        convert_to_dependency(src, dst, language='zh', version='4.2.0')
    else:
        raise

When it happens

Trigger: Running make_ctb_tasks/make_dep_conllx on a machine without java on PATH; java too old/new for the Stanford Parser jar; large treebank conversions exhausting heap (default JVM max heap).

Common situations: Fresh CI containers or slim Docker images without a JDK; converting full CTB9 where the parser needs several GB of heap.

Related errors

• Unsupported version {version}
• Got average f{average}, expected one of None, 'token', or 'b
• alpha must be float, list of float, or torch.FloatTensor, {}
• Only supports floating point dtypes.
• Does not support dtype " + str(dtype)

AI-assisted analysis of hankcs/HanLP@ddb1299bdd (2026-08-27). Data as JSON: /api/errors/13d7954d1a74dc22. Report an issue: GitHub.