hanzi2reading v0.0.2

hanzi2reading

This library is distributed with a default database; you can find the license information here. https://github.com/g0v/moedict-data/blob/master/README.md

Design goals

• Annotation of Chinese characters with Standard Mandarin (國語/普通話) readings
• Agnostic to simplified/traditional script and transliteration method
• Should work offline, and database format should be as compact as possible - e.g. Protocol Buffers loaded by WebAssembly
• Should support word-based disambiguation of characters with multiple readings
• separation of code and data - dictionary backend should be swappable

Limitations

• Word segmentation is a non-goal.
• Target should be good performance for non-sentence inputs, without needing part-of-speech classification, e.g. 得

Database Format

• https://en.wikipedia.org/wiki/Zhuyin_table
• https://en.wikipedia.org/wiki/Pinyin_table

Total = 15 bits per syllable. This is less compact than enumerating all standard syllables, but allows dictionaries to have non-standard syllables.

Notes

• https://www.unicode.org/reports/tr38/#N1019C
• https://languagelog.ldc.upenn.edu/nll/?p=45843

Resources

• https://github.com/mozillazg/python-pinyin (SC only, data embedded in code)
• https://github.com/tsroten/dragonmapper (data is in large CSV files, Python only)
• https://github.com/g0v/moedict-data
• https://cc-cedict.org/editor/editor.php
• https://chrome.google.com/webstore/detail/zhongwen-chinese-english/kkmlkkjojmombglmlpbpapmhcaljjkde
• https://github.com/skishore/makemeahanzi