Corpus & languages

Segments live under data/corpus/<lang>/segments.jsonl.

Primary languages (≥1000 segments each):

Code	Language
en	English
zh	Chinese (zh-Hans + zh-Hant treated as Chinese)
ja	Japanese
es	Spanish
fr	French

Other detected languages fall back to English.

Regenerate:

uv run python scripts/generate_corpus.py

Custom Q&A

Edit data/corpus/custom_qa.json:

[
  {
    "pattern": "(?i)\\bping\\b",
    "match": "regex",
    "reply": "pong",
    "langs": ["*"]
  }
]

match: regex | contains | exact.