Parse and convert between different Teochew phonetic spelling schemes.
Project description
Parsing and conversion of Teochew Chinese romanizations
Parse and convert between different Teochew romanized spelling schemes.
parsetc
represents romanization schemes as context-free grammars, and
implements parsers and translators using the
lark
Python
library. The aim is to represent romanized text as an abstract parse tree
representing the phonology, which can then be converted to other romanization
schemes in a consistent, rule-based way.
Available romanization schemes for Teochew
Input
- Geng'dang Pêng'im 廣東拼音 (
gdpi
) - Gaginang Peng'im 家己儂拼音 (
ggn
) - Gaginang Peng'im with coda
-n
allowed (nasalization written withñ
instead) (ggnn
) - Dieghv 潮語 (
dieghv
); (source) - Tie-tsiann-hue 潮正會, also known as Tie-lo 潮羅 (
tlo
); (source) - Duffus system (
duffus
), also known as Pe̍h-ūe-jī (PUJ); (source)
Output
gdpi
, ggnn
, tlo
, duffus
, dieghv
, plus:
- Teochew Sinwenz (
sinwz
); (source) - Traditional initial-final categories (
15
) from 《彙集雅俗通十五音》 also known as 《擊木知音》, based on the analysis by 徐宇航 「《擊木知音》音系之再研究」 (2014)
Other languages
This tool was originally developed for Teochew, but can be extended in the
future to other dialects or languages. An experimental Cantonese module
(--language Cantonese
) with Jyutping (jp
) and Cantonese Pinyin (cpy
) is
available.
Orthographic requirements for input text
- Text must be in lower case, otherwise will be converted to lower case
- Syllables may be written with or without tone numbers
- Syllables may be combined into words for legibility, i.e. word-segmented writing
- If syllables are combined into words, they must either have tone numbers
(e.g.
diê5ziu1
), or use a syllable separator character (e.g.diê-ziu
orpêng'im
, ordiê5-ziu1
). The separator character is either a hyphen or single apostrophe. This is because of ambiguous parsings, e.g.pê-ngi-m
instead ofpêng-im
, which in general can only be dealt with by usage frequency, which is not available. - If syllable separator characters are used, they must be used consistently. Mixing conventions may cause unexpected parsing errors.
Installation
parsetc
requires Python 3 and lark
v1.1.
Install latest release with pip
from PyPI:
pip install parsetc
If you are interested in latest development version, you can clone this
repository and checkout the dev
branch, then install with pip
from source
code:
pip install .
See help message:
parsetc --help
View available input and output schemes for Teochew:
parsetc --language Teochew --show_options
Usage
Command line tool
Input text is read line-by-line from STDIN. Output is written to STDOUT.
The language (--language
or -l
) is Teochew
by default.
# Convert to Tie-lo
echo 'ua2 ain3 oh8 diê5ghe2, ain3 dan3 diê5ziu1 uê7.' | parsetc -i gdpi -o tlo
# Convert to all available output romanizations
echo 'ua2 ain3 oh8 diê5ghe2, ain3 dan3 diê5ziu1 uê7.' | parsetc -i gdpi --all
# Show parse tree (useful for debugging)
echo 'ua2 ain3 oh8 diê5ghe2, ain3 dan3 diê5ziu1 uê7.' | parsetc -i gdpi -p
Testing with sample texts in the examples/
folder:
# Example with tone numbers but no syllable separators
cat examples/teochew.dieghv.tones.txt | parsetc -i dieghv --all
# Example with hyphens as syllable separators
cat examples/teochew.dieghv.sep.txt | parsetc -i dieghv --all
Try a Cantonese example (work in progress):
echo "ceon1 min4 bat1 gok3 hiu2" | parsetc -l Cantonese -i jp --all
cat examples/cantonese.cpy.txt | parsetc -l Cantonese -i cpy --all
Python module
[ Work in progress. Updated docs coming soon... ]
Common functions and command line script are in the parsetc.parsetc
submodule. Each language has a dedicated subpackage with the following
structure (using Teochew as an example):
src/parsetc/Teochew/
├── shared.lark # lark grammar for shared parser rules
├── extends.json # lark %extend statements specific to individual schemes
├── terminals.json # dictionary of terminals for each scheme
├── mergers.json # dictionary of phonological mergers specific to individual schemes
├── parser.py # preprocessing and parser functions
└── translit.py # Translator classes
The lark rules and json files are the data required for the parsing and
translation functions in parser.py
and translit.py
.
Q & A
-
Q: Can I convert Chinese characters to romanization with
parsetc
? -
A: No, it is for converting between different romanization systems only.
-
Q:
parsetc
makes a mistake, or it crashes when parsing a text. Can you fix it? -
A: Please check first if you are using the latest version of
parsetc
(runparsetc --version
). If the error still persists, please file a bug report on the Github issues page, with the input text that caused the crash/error. -
Q: Can I contribute code to this project?
-
A: You are welcome to look into the source code and fork/modfiy it. However, I currently do not have the capacity to review pull requests and contributions in detail, so please understand if I decline them or take a long time to respond.
Related projects
From Learn-Teochew
We use parsetc
to convert Teochew opera transcriptions into different
romanization systems for the Learn Teochew
Opera website.
How to type the special diacritics like ṳ, o̍, o͘
in some romanization systems?
If you use MacOS X, check out our custom keyboard
layouts
Others
- taibun - Transliterator and tokenizer for Taiwanese Hokkien by Andrei Harbachov
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file parsetc-0.4.1.tar.gz
.
File metadata
- Download URL: parsetc-0.4.1.tar.gz
- Upload date:
- Size: 20.8 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/4.0.2 CPython/3.9.7
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | ecd1cfd99dd88080847fa86aa93ba10bff47007c9aa7d9450e0a6ce5c5d87a0b |
|
MD5 | 134266620745490a482794e8e2b64f07 |
|
BLAKE2b-256 | 06d8bb7aa7ceb33033dd26a6d13acf140ee58efef969e989c7d03511aeef8395 |
File details
Details for the file parsetc-0.4.1-py3-none-any.whl
.
File metadata
- Download URL: parsetc-0.4.1-py3-none-any.whl
- Upload date:
- Size: 22.1 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/4.0.2 CPython/3.9.7
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 8ae01b1898dd5737aa71827df40b36e7c1279a724c77463c5b33a4135f4615ae |
|
MD5 | f73ac20cd072b4c2df447535fed90f5a |
|
BLAKE2b-256 | ec8878cb78fccd9f58cd8f69657fa1dab48ad29567eb7975d87bc9fce63e0b8c |