keytao-lookup

Call keytao_lookup_by_word with the Chinese word to get all its codes from the dictionary, or keytao_encode to compute codes by rule for words not in the dictionary. Batch variants accept up to 100 words per call.

Use keytao_encode, which returns per-character pinyin, phonetic codes, character roots, and shape codes computed by encoding rules. The database lookup tools do not contain split information and cannot substitute for it.

Yes, keytao_lookup_by_words_batch and keytao_lookup_by_codes_batch each accept up to 100 items per call. Single-item queries should use keytao_lookup_by_word or keytao_lookup_by_code instead.

It calls a Keytao Next API instance configured via KEYTAO_API_BASE, defaulting to https://keytao.vercel.app, and requires a bot API token from the NoneBot config. Requests use httpx with retry logic and 10-20 second timeouts.

Encoding fails when the service cannot determine pronunciation or the word contains unsupported characters, returning codes with question marks or an empty list. In that case the user must manually specify a code through the change-code flow.

Yes, keytao_word_commonness compares 1 to 12 words using local BCC corpus frequencies and dictionary evidence without modifying the dictionary. It reports per-word frequency, dictionary inclusion, and ties explicitly rather than treating missing data as zero.