rongzi

Input Chinese characters, and the application will find optimized component-paths connecting them. It's like melting a word. Or traversing a banyan grove by following branches between prop-trunks.
With appreciation and respect for the late poet Yungtze (1922-2021).

input: 蓉 榮

output: 蓉🡒容🡒榕🡒木🡒榮

---

Link to streamlit demonstration webapp:

http://rongzi.bhrdwj.net

Usage:

Simply connect the characters of a poem:

    import rongzi
    rz = rongzi.RongZi()
    df = rz.analyze_sequence('永遠的青鳥')
    print(df.to_markdown())

	永遠	遠的	的青	青鳥
0	永栐木榬袁遠	遠薳艹菂的	的白皘青	青鶄鳥
1	永水閖門闧遠	遠辶迫白的	的白胉月青	青聙耳鵈鳥
2	永詠言誾門闧遠	遠袁鎱金鉑白的	的白日晴青	青錆金鵭鳥
3	永詠言這辶遠	遠袁褤衤袙白的	的白鲌鱼鲭青	青靖立鴗鳥
4	永泳氵溒袁遠	遠袁𠮷口啲的	的白鉑金錆青	青鯖魚鷠鳥

Installation and dependencies:

• For basic usage rongzi requires:
  • python version 3.9 or newer.
  • Third-party libraries pandas (1.3.4 or newer), and numpy.
• For graph visualization rongzi additionally needs:
  • graphviz
  • (Sometimes graphviz may be a trickier install, so I split this functionality out into a separate library, starting with the streamlit app app.py.)

Challenge specification:

• Query the CCD database of Chinese character radical breakdowns,
  • modeled as a directed graph.
• Develop an process that can:
  • when given a Chinese poem as an input,
  • find an optimized path stringing the characters together,
  • by moving from character to character via shared components and intermediate characters,
  • while:
    • using the fewest possible steps,
    • avoiding using the basic and compound strokes as components to link characters
    • with minimal duplication of characters.
• Present it as a hosted interactive web-app.

The name "rongzi"

熔字，róngzì
蓉子，Róngzi
榕子，róngzi
榮。róng

Molten words,
Yungtze,
a banyan seed,
and honor.

TODO:

• Refactor (ccd, pdb, kdb) as class property methods:
  • https://stackoverflow.com/questions/128573/using-property-on-classmethods
• Allow to choose between CJK languages/scripts (CH, JA, TW)
• Create CICD pipeline using github actions.
• Dockerize application