kre
Subcharacter-level regular expressions with Korean text.
kre is a wrapper for re
from the Python Standard Library which allows users to apply the full functionality of re
at the subcharacter level for Korean text.
Installation
kre releases are available on PyPI.
pip install kre
Documentation
Most functionality is documented in the re documentation.
Documentation on the unique features of kre is available in the wiki, where you will also find discussion of inherent differences between re
(character-level regular expressions) and kre
(subcharacter-level regular expressions) and how kre addresses them. It is strongly recommended that users familiarize themselves with these differences.
Example Features
In the simple case of search functions, matches are mapped back to their original position.
> re.search(r"γ
‘", "νκΈ") # no match
> kre.search(r"γ
‘", "νκΈ")
<kre.KRE_Match object; span=(1, 2), match='κΈ'>
In the case of subcharacter-level substitutions, kre can recombine any newly created sequences into standard Korean characters, provided the input used standard (syllable) characters.
> kre.sub(r"γ
", r"γ
", "ν³γ
νγ
γ
νν³")
'νγ
νΈγ
γ
νΈν'
If you prefer, kre can also attempt to merge non-standard input with substitutions.
> kre.sub(r"γ
", r"γ
", "ν³γ
νγ
γ
νν³", syllabify="extended")
'νΈνΈνΈνΈνΈν'
Although linearizing a Korean string normally results in the loss of information about syllable boundaries, kre makes it possible to make use of syllable boundaries in regular expression patterns through the use of (customizable) syllable delimiters (';' by default).
> kre.search(r"γ
", "μμΌ μΆνν΄~")
<kre.KRE_Match object; span=(0, 1), match='μ'>
> kre.search(r";γ
", "μμΌ μΆνν΄~", boundaries=True)
<kre.KRE_Match object; span=(1, 2), match='μΌ'>
As a more interesting, complicated, and perhaps useless example of what kre can do, the following swaps every sequential pair of final consonant(s) (λ°μΉ¨) in the input string.
> sun_and_moon = "μλ μμ κΉμ μ° μμ κ°λνμ§λ§ μ¬μ΄μ’μ μ€λμ΄μ κ·Έ νμ΄λ¨Έλ κ°μ‘±μ΄ μ΄κ³ μμλ€."
> kre.sub(r"([γ
-γ
£])([γ±-γ
]{1,2};)(.*?)([γ
-γ
£])([γ±-γ
]{1,2};)", r"\1\5\3\4\2", sun_and_moon, boundaries=True)
'μλ« μμ κΈ΄μ μ μμ κ°λνμ§λ§ μ¬μ΄μ‘΄μ μ€λμ΄μ κ·Έ νΉμ΄λ¨Έλ κ°μ‘Έμ΄ μκ³ μΌμλ€.'