o i&@s*dZddlZddlmZddlmZddlZddlZej ej e Z dZ dZdZdZeeje d d d  ZeeZWdn1sGwYd d eDZeeje d dd  ZeeZWdn1spwYdd eDZddeddDZddeddDZddeddDZddeddDZddeddDZddeddDZ Gdd d e!Z"d!d"Z#dId#d$Z$d%d&Z%d'd(Z&d)d*Z'd+d,Z(d-d.Z)d/d0Z*d1d2Z+d3d4Z,d5d6Z-d7d8Z.dJd:d;Z/dJdd?Z1d@dAZ2dKdCdDZ3dIdEdFZ4dGdHZ5dS)La$Syllable and jamo analysis for Korean. Default internal exchange form is Hangul characters, not codepoints. Jamo exchange form is U+11xx characters, not U+3xxx Hangul Compatibility Jamo (HCJ) characters or codepoints. For more information, see: http://python-jamo.readthedocs.org/ko/latest/ N)stderr)chaini`idataz U+11xx.jsonrcCi|]\}}||qSr .0charnamer r =/home/ubuntu/.local/lib/python3.10/site-packages/jamo/jamo.py rz U+31xx.jsoncCrr r r r r rrrcCg|]}t|qSr chrr _r r r r_cCrr rrr r rr ricCrr rrr r rr!racCrr rrr r rr"rivcCrr rrr r rr#ricCrr rrr r rr$rics eZdZdZfddZZS)InvalidJamoErrorzjamo is a U+11xx codepoint.cs@tt||tt||_tdj|jdddtddS)NzCould not parse jamo: U+{code})code)file) superr__init__hexordjamoprintformatr)selfmessager$ __class__r rr!)s  zInvalidJamoError.__init__)__name__ __module__ __qualname____doc__r! __classcell__r r r)rr'srcCs~t|r=t|t}|d}d||dd}d|d}|r1t|tt|tt|tfSt|tt|tfS|S)zpReturn a 3-tuple of lead, vowel, and tail jamo characters. Note: Non-Hangul characters are echoed back. L)is_hangul_charr# _JAMO_OFFSETr_JAMO_LEAD_OFFSET_JAMO_VOWEL_OFFSET_JAMO_TAIL_OFFSET)syllableremtailvowelleadr r r_hangul_char_to_jamo0s       r=cCsPt|t}t|t}|rt|tnd}t||dd|ddtS)z?Return the Hangul character for the given jamo characters. rr1r0r2)r#r5r6r7rr4r<r;r:r r r_jamo_to_hangul_charDs  $r?cCs2t|rtddt|}|tvrt|S|S)N(?<=HANGUL )(\w+)LETTER)is_jamoresub_get_unicode_name_HCJ_REVERSE_LOOKUPkeys)r hcj_namer r r_jamo_char_to_hcjMs rIcCs:|tvr|tvrtd|t|rt|St|S)z0Fetch the unicode name for jamo characters. z#Not jamo or nameless jamo character) _JAMO_TO_NAMErG _HCJ_TO_NAMEris_hcj)r r r rrEWs  rEcCspt|}d|ko dknp7d|kodknp7d|ko%dknp7d|ko1dknp7t|S) zTest if a single character is a jamo character. Valid jamo includes all modern and archaic jamo, as well as all HCJ. Non-assigned code points are invalid. rii`i|iiii)r#rL characterrr r rrBbsrBcCsXt|}d|ko dknp+d|kodknp+d|ko%dknp+t|S)a*Test if a single character is a modern jamo character. Modern jamo includes all U+11xx jamo in addition to HCJ in modern usage, as defined in Unicode 7.0. WARNING: U+1160 is NOT considered a modern jamo character, but it is listed under 'Medial Vowels' in the Unicode 7.0 spec. ririuri)r# is_hcj_modernrMr r ris_jamo_modernnsrPcCs(dt|ko dknot|dkS)zTest if a single character is a HCJ character. HCJ is defined as the U+313x to U+318x block, sans two non-assigned code points. 11i1id1r#rNr r rrL|s(rLcCs8t|}d|ko dknpd|kodkSS)zTest if a single character is a modern HCJ character. Modern HCJ is defined as HCJ that corresponds to a U+11xx jamo character in modern usage. rQiN1O1c1rRrMr r rrOs rOcCsdt|ko dkSS)zfTest if a single character is in the U+AC00 to U+D7A3 code block, excluding unassigned codes. rirRrSr r rr3sr3cCsd|tvs |tdkr dS|tvs%|tdks%dt|kr#dkr'dSndS|tvr-dStd|) aDetermine if a jamo character is a lead, vowel, or tail. Integers and U+11xx characters are valid arguments. HCJ consonants are not valid here. get_jamo_class should return the class ["lead" | "vowel" | "tail"] of a given character or integer. Note: jamo class directly corresponds to the Unicode 7.0 specification, thus includes filler characters as having a class. rr<rrTrUr;r:z#Invalid or classless jamo argument.) JAMO_LEADSr JAMO_VOWELSr# JAMO_TAILSrr$r r rget_jamo_classs  rZcCsdd|DS)aYConvert jamo to HCJ. Arguments may be iterables or single characters. jamo_to_hcj should convert every jamo character into HCJ in a given input, if possible. Anything else is unchanged. jamo_to_hcj is the generator version of j2hcj, the string version. Passing a character to jamo_to_hcj will still return a generator. cs|]}t|VqdSN)rIrr r r zjamo_to_hcj..r )rr r r jamo_to_hcj r_cCdt|S)aConvert jamo into HCJ. Arguments may be iterables or single characters. j2hcj should convert every jamo character into HCJ in a given input, if possible. Anything else is unchanged. j2hcj is the string version of jamo_to_hcj, the generator version. )joinr_rYr r rj2hcjs rdr;cCs^|dkrd}n|dkrd}n |dkrd}ntd|td|t|}|tvr-t|S|S) zConvert a HCJ character to a jamo character. Arguments may be single characters along with the desired jamo class (lead, vowel, tail). Non-mappable input will raise an InvalidJamoError. r<CHOSEONGr; JUNGSEONGr: JONGSEONGzNo mapping from input to jamo.r@)rrCrDrE_JAMO_REVERSE_LOOKUPrG)hcj_charposition jamo_class jamo_namer r r hcj_to_jamos  rmcCs t||S)zOConvert a HCJ character to a jamo character. Identical to hcj_to_jamo. )rm)rirjr r rhcj2js rncCsddtdd|DDS)a4Convert a string of Hangul to jamo. Arguments may be iterables of characters. hangul_to_jamo should split every Hangul character into U+11xx jamo characters for any given string. Non-hangul characters are not changed. hangul_to_jamo is the generator version of h2j, the string version. css|]}|VqdSr\r rr r rr]sz!hangul_to_jamo..csr[r\)r=rr r rr]r^)r from_iterable hangul_stringr r rhangul_to_jamos  rrcCra)aConvert a string of Hangul to jamo. Arguments may be iterables of characters. h2j should split every Hangul character into U+11xx jamo for any given string. Non-hangul characters are not touched. h2j is the string version of hangul_to_jamo, the generator version. rb)rcrrrpr r rh2jr`rsrbcCst|d}t|d}|rt|dkrd}n t|rt|d}t|rJt|dkrJt|rJt|dkrJ|r>t|rJt|dkrJt|||}t|rJ|Stdd)zReturn the Hangul character for the given jamo input. Integers corresponding to U+11xx jamo codepoints, U+11xx jamo characters, or HCJ are valid inputs. Outputs a one-character Hangul string. This function is identical to j2h. r<r;rNr:z*Could not synthesize characters to Hangul.)rmr#rLrBrZr?r3r)r<r;r:resultr r rjamo_to_hanguls*     rvcCs t|||S)zArguments may be integers corresponding to the U+11xx codepoints, the actual U+11xx jamo characters, or HCJ. Outputs a one-character Hangul string. This function is defined solely for naming conisistency with jamo_to_hangul. )rvr>r r rj2hs rwcCst)zAConvert jamo characters in a string into hcj as much as possible.)NotImplementedErrorrc)stringr r r synth_hangul&srz)r)r;)rb)6r.ossysr itertoolsrjsonrCpathabspathdirname__file___ROOTr4r5r6r7openrcnamedataloadrJitemsrhrKrFrangerVJAMO_LEADS_MODERNrWJAMO_VOWELS_MODERNrXJAMO_TAILS_MODERN Exceptionrr=r?rIrErBrPrLrOr3rZr_rdrmrnrrrsrvrwrzr r r rsZ