j<3 6UdZddlmZddlZddlZgdZded<iddd d d d d ddddddddddddddddddd d!d"d#d$d%d&d'd(id)d*d+d,d-d.d/d0d1d2d3d4d5d6d7d8d9d:d;d<d=d>d?d@dAdBdCdDdEdFdGdHddIidJdKddLddMdNdOdPdQd,dRdSdTdUdVdWdXd*dYdZd[d@d\d4d]d^d_dBd`dadbdIdcidddedfdgdKdhdQdidjdkdldkdmdndMdodpdqdrdsdedtdududcdvdwdxdydzd{d|d}d~dddddd;dZded<ejd de e e dDZ gdZgdZded<iZded<eD]"\ZZedD]Zeee<#ejd de e e dDZgdZGddZeZddZeZddlmZmZGddeZGddeeefZ dS)aVietnamese CAD Font Normalizer and Multi-Encoding Transcoder. Supports TCVN3 (ABC), VNI-Windows, AutoCAD SHX/MText Escape Sequences, and Unicode Normalization. Specialists: Quynh (QS & BoQ Takeoff), Hung (Site Engineering & Drawings), Minh (System Standards). ) annotationsN))u¦íướ)u¦êườ)u¦ëưở)u¦ìưỡ)u¦îượ)u¦¬ươ)u­ír)u­êr)u­ër)u­ìr)u­îr)u­¬r )uªÕế)uªÒề)uªÓể)uªÔễ)uªÖệ)u«èố)u«åồ)u«æổ)u«çỗ)u«éộzlist[tuple[str, str]]TCVN3_COMPOSITES¸áµà¶ả·ã¹ạ¨ă¾ắ»ằ¼ẳ½ẵÆặ©âÊấÇầÈẩÉẫËậÐéÌèÎẻÏẽÑẹªêÕr Òr Ór Ôr ÖrÝí×ìØỉÜĩÞịóßòỏõäọ«ôrårærçrr¬ơớờëởỡîợúïùñủũụ­ư¦øứừöử÷ữựýỳûỷüỹþuỵ®đĐĂÂ)§¡¢£¤¥zdict[str, str] TCVN3_DICT|c#>K|]}tj|VdSNreescape.0ks TC:\Projects\DSCons\app\modules\takeoff\application\vietnamese_cad_font_transcoder.py r~s* T TaRYq\\ T T T T T TT)keyreverse)8rrr!r%r)r-r/r7rCrGrDrErFrHrrrrrrrrprr+rYr^rnr#r'r5r9r;r=r?rJrLrNrPrr[r\r]r:r8rBrbrKrIrerjrTrZrVrtrvrq))u oâù|oár)u oâø|oàr)u oâû|oår)u oâõ|oãr)u oâï|oär)oârZ)u OÂÙ|OÁuỐ)u OÂØ|OÀuỒ)u OÂÛ|OÅuỔ)u OÂÕ|OÃuỖ)u OÂÄ|OÄuỘ)uOÂrF)u aâù|aár.)u aâø|aàr0)u aâû|aår2)u aâõ|aãr4)u aâï|aär6)uaâr,)u AÂÙ|AÁuẤ)u AÂØ|AÀuẦ)u AÂÛ|AÅuẨ)u AÂÕ|AÃuẪ)u AÂÄ|AÄuẬ)uAÂr)u aêù|aér")u aêø|aèr$)u aêû|aúr&)u aêõ|aür()u aêï|aër*)aêr )u AÊÙ|AÉuẮ)u AÊØ|AÈuẰ)u AÊÛ|AÚuẲ)u AÊÕ|AÜuẴ)u AÊÄ|AËuẶ)uAÊr)u eâù|eár )u eâø|eàr )u eâû|eår )u eâõ|eãr )u eâï|eär)eârB)u EÂÙ|EÁuẾ)u EÂØ|EÀuỀ)u EÂÛ|EÅuỂ)u EÂÕ|EÃuỄ)u EÂÄ|EÄuỆ)uEÂr-)ôùr`)ôøra)uôûrc)uôõrd)uôïrf)rZr_)uÔÙuỚ)uÔØuỜ)uÔÛuỞ)uÔÕuỠ)uÔÏuỢ)rFuƠ)öùrr)öørs)uöûru)uöõrw)uöïrx)rtro)uÖÙuỨ)uÖØuỪ)uÖÛuỬ)uÖÕuỮ)uÖÏuỰ)rGuƯ)aùr)aør)aûr)aõr)aïr)uAÙÁ)uAØÀ)uAÛuẢ)uAÕÃ)uAÏuẠ)eùr8)eør:)eûr<)eõr>)eïr@)uEÙr3)uEØr1)uEÛuẺ)uEÕuẼ)uEÏuẸ)oùrR)oørT)oûrU)oõrV)oïrX)uOÙrE)uOØrD)uOÛuỎ)uOÕrC)uOÏuỌ)uùrg)uøri)uûrk)uõrl)uïrm)uUÙÚ)uUØÙ)uUÛuỦ)uUÕuŨ)uUÏuỤ)yùry)yørz)uyûr|)uyõr~)uYÙrH)uYØuỲ)uYÛuỶ)uYÕuỸ)rjr)r?r)rIrI)rKrK)r\rM)rRrO)rTrQ)Ír)r9r9)r)uỈ)rEuĨ)rDuỊ VNI_PATTERNS _VNI_DICTc#>K|]}tj|VdSrrrs rrrPs* S SaRYq\\ S S S S S Sr)rrrrrrrrrrrrrrrrrrrrrrrrrrrrrrjr?ceZdZdZed dZed dZed dZedd Z ed d Z d d Z d S)VietnameseCadFontTranscoderzkSpecialized transcoder to decode all legacy Vietnamese CAD fonts (TCVN3, VNI, SHX, MText) to UTF-8 Unicode.textstrreturncr|sdSdd}tjd||}tjdd |}tjd d |}tjd d |}tjdd|}tjdd|}tjdd|}tjdd|}tjdd|}tjdd|}tjdd|}|dddd}|S)zIStrip AutoCAD MText formatting codes, escape sequences, and symbol codes.matchre.Matchrrc|d} tt|dS#t$r|dcYSwxYw)Nr)groupchrint Exception)rhex_vals r_replace_u_hexzGVietnameseCadFontTranscoder.clean_autocad_mtext.._replace_u_hexs_kk!nnG &3w++,,, & & &{{1~~%%% &s4AAz(?i)\\U\+([0-9a-f]{4})z(?i)%%crLz(?i)%%d°z(?i)%%p±z(?i)%%uz(?i)%%oz(?i)%%%%z(?i)\\P z\\~ z(?i)\\[fFHhWwQqAaCc]\w*[^;]*;z(?i)\\S([^;]*);z\1{})rrrr)rsubreplacestrip)rrress rclean_autocad_mtextz/VietnameseCadFontTranscoder.clean_autocad_mtextys2 2 & & & &f.EEfZs++fZs++fZs++fZS))fZS))fZc**fZs++fVS#&&f5r3??f'44kk#r""**333yy{{rc|sdStD]\}}|||}td|}t jd|S)z@Convert a TCVN3 (ABC / .VnTime) encoded string to UTF-8 Unicode.rct|d|dSNr)rgetrms rzHVietnameseCadFontTranscoder.transcode_tcvn3_to_unicode..s( AGGAJJ)O)OrNFC)rr _TCVN3_REGEXr unicodedata normalize)clsrcompreplrs rtranscode_tcvn3_to_unicodez6VietnameseCadFontTranscoder.transcode_tcvn3_to_unicodesf 2* , ,JD$<<d++DDOOQUVV$UC000rcl|sdStd|}tjd|S)z6Convert a VNI-Windows encoded string to UTF-8 Unicode.rcBt|dSr)rrrs rrzFVietnameseCadFontTranscoder.transcode_vni_to_unicode..sy'<rr) _VNI_REGEXrrr)rrrs rtranscode_vni_to_unicodez4VietnameseCadFontTranscoder.transcode_vni_to_unicodes; 2nn<.s/==Q a((======rc3BK|]}|VdSrr)rrrs rrzFVietnameseCadFontTranscoder.auto_transcode_cad_text..s/RRq'--**RRRRRRrr) rrrrsum _VNI_PAIRS_TCVN3_DISCRIMINATIVE_MARKERSrr)rr vni_score tcvn3_scorers @rauto_transcode_cad_textz3VietnameseCadFontTranscoder.auto_transcode_cad_texts 2))$//  . .w 7 7 9(88 8====*===== RRRR4QRRRRR q==Y44//88 8 1__11':: : ]]//88 8$UG444rc,||S)z+Instance alias for auto_transcode_cad_text.)r)selfrs rdecode_cad_textz+VietnameseCadFontTranscoder.decode_cad_texts++D111rN)rrrr)rrrr) __name__ __module__ __qualname____doc__ staticmethodr classmethodrrrrrrrrrvsuu"""\"H111[1111[1["555[54222222rrrctSr)_global_transcoderr rrget_cad_font_transcoderrs r)IQuery IQueryHandlerceZdZUdZded<dS)DecodeCadTextQueryz]CQRS Query to decode AutoCAD text with legacy Vietnamese fonts (TCVN3, VNI) to UTF-8 Unicode.rraw_textN)rrr r __annotations__r rrrrsggMMMMMrrceZdZddZdS)DecodeCadTextQueryHandlerqueryrrrcRt}||jSr)rrr)rr transcoders rhandlez DecodeCadTextQueryHandler.handles#,.. 11%.AAArN)rrrr)rrr rr rrrrs.BBBBBBrr)rr)!r  __future__rrrrrrcompilejoinsortedkeyslenrrrrp_grouprsplitsub_prrrrr"get_vietnamese_cad_font_transcoderapp.core.module_framework.cqrsrrrrrr rrr(s #"""""  +++2S DS DS E S  D S  E S DS ES ES ES ES ES DS ES ES  E!S" E#S$ E%SS( D)S* D+S, E-S. E/S0 E1S2 D3S4 E5S6 E7S8 E9S: E;S< E=S@ DASB DCSD EESF DGSH EISL DMSSSN DOSP EQSR DSST EUSV DWSX EYSZ E[S\ E]S^ E_S` EaSb DcSd EeSf EgSh EiSj EkSl EmSp DqSSSr DsSt EuSv DwSx EySz D{S| D}S~ ES@ EASB ECSD EESF EGSJ DKSL EMSN EOSP EQSR ESSV DWSSX      eSSS SSSSjrzHH T T66*//*;*;d#S#S#S T T TTT 9!9!9!~H'H'H' HHHHT !  MGTs##   % RZHH S S66)..*:*:T#R#R#R S S SSS     Fi2i2i2i2i2i2i2i2X1022&=" A@@@@@@@BBBBB .@#.E FBBBBBr