Tojyutping

What's Changed
- **Update Dictionary Data**: Slightly reduced dictionary size.

**Full Changelog**: https://github.com/CanCLID/ToJyutping/compare/3.1.0...3.2.0

What's Changed
* Changes to the `g2p` method:
* We removed the restriction that did not allow patching `unknown_id` without supplying `puncts_map` since the built-in punctuation mapping can already be patched by the `extra_puncts` option.
* **Fixed:** The `lengths` attribute of the output `PhonemesList` now agrees with the original input. Each element of `lengths` is now the number of elements of `segmentals` or `tones` that correspond to each character of the input instead of each element of the `PhonemesList`.

**Full Changelog**: https://github.com/CanCLID/ToJyutping/compare/3.0.0...3.1.0

What's Changed
* **Breaking Change**: Internal methods are no longer exposed by the default entrypoint.
* Changes to the `g2p` method:
* **Breaking Change**: The output list now includes fillers for unknown characters (1), punctuations (from 2 to 7) in addition to syllable components (from 8 to 94). Unknown character fillers and punctuations are output as singletons (1-tuples). The values can be adjusted by the `offset` and `puncts_offset` arguments.
* The output list now contains useful properties, namely `segmentals`, `tones` and `lengths`.
* Punctuations can be added by the `extra_puncts` argument or customized by the `puncts_map` + `unknown_id` arguments.
* Read the [documentation](../../grapheme-to-phoneme-conversion-function) for more info.
* The new `customize` method:
* Adds the ability to include custom entries and override or exclude built-in entries.
* The constructed converters can be chained without affecting each other.
* Read the [documentation](../../custom-entries--existing-entries-overriding-or-exclusion) for more info.
* The `get_jyutping` and `get_ipa` methods are slightly optimized.

**Full Changelog**: https://github.com/CanCLID/ToJyutping/compare/2.0.0...3.0.0

What's Changed
* **Breaking Change**: `g2p` now outputs tones as integers from 1 to 6 (instead of 87 to 92) by default
* To retain the old behavior, pass the argument `tone_same_seq=True`
* Accept a triplet as the `offset` argument of `g2p`
* Slightly optimize the two `get_*_candidates` methods

**Full Changelog**: https://github.com/CanCLID/ToJyutping/compare/1.0.0...2.0.0

What's Changed
* Completely rewrite the codebase in an object-oriented manner
* Add `g2p` (grapheme-to-phoneme) conversion function for machine learning purposes
* Optimize performance & memory usage (https://github.com/CanCLID/ToJyutping/pull/7)

**Full Changelog**: https://github.com/CanCLID/ToJyutping/compare/0.3.0...1.0.0

What's Changed
- Use new dictionary to increase accuracy and reduce module size
- Add methods for retrieving all possible pronunciations of a character
- Drop dependency
- Drop Python 3.5, Add 3.11 & 3.12

**Full Changelog**: https://github.com/CanCLID/ToJyutping/compare/0.2.3...0.3.0