haqumei-cli

A command-line interface for haqumei, a G2P (Grapheme-to-Phoneme) and text processing library for Japanese.

Installation

cargo install haqumei-cli

Usage

REPL

Run without arguments to enter the interactive mode.

$ haqumei-cli
Enter text to process (Ctrl+C or Ctrl+D to exit):
> 今日はいい天気ですね。
ky o o w a i i t e N k i d e s U n e

One-liner / Pipeline Processing

You can specify the text directly as an argument, or pass the text via standard input.

$ haqumei-cli "吾輩は猫である" --mode kana-per-word
ワガハイ ワ ネコ デ アル

$ echo "吾輩は猫である" | haqumei-cli --mode kana-per-word
ワガハイ ワ ネコ デ アル

File Processing

Read from an input file and write to an output file.

haqumei-cli --input input.txt --output output.txt --mode g2p

JSON Lines Output

Supports structured JSON output.

$ haqumei-cli "テスト" --mode mapping-detailed --format json
[{"word":"テスト","phonemes":["t","e","s","U","t","o"],"features":["テスト","名詞","サ変接続","*","*","*","*","テスト","テスト","テスト","1/3","C1"],"pos":"名詞","pos_group1":"サ変接続","pos_group2":"*","pos_group3":"*","ctype":"*","cform":"*","orig":"テスト","read":"テスト","pron":"テス’ト","accent_nucleus":1,"mora_count":3,"chain_rule":"C1","chain_flag":-1,"is_unknown":false,"is_ignored":false}]

Modes (--mode)

haqumei-cli supports various output modes:

• g2p (default): Flat phoneme sequence.
• g2p-detailed: Detailed phoneme sequence (symbols converted to sp, unk, etc.).
• kana: Katakana sequence.
• kana-per-word: Katakana sequence separated by word.
• per-word: Phoneme list grouped by word.
• pairs: Word to phoneme mapping.
• mapping: Detailed mapping including unknown word status.
• mapping-detailed: Includes POS, pronunciation, accent nucleus, and mora count.
• fullcontext: Full-context labels for TTS.

Options

Run --help to see all available configuration options (e.g., dictionary paths, unicode normalization, long vowel rules).

haqumei-cli --help