j36ddlZddlZ edZiddddddddd d d d d ddddddddddddddddddd d!d"d#d$d%d&d'd(d)d*d+d,d-d.d/d0d1Zd2d3d3d4Ziid5d2d6d7d8d9d:d;dd?d@dAdBdCdDdEdFdGdHdIdJdKdLd2dMdNdOdNdPdQd2d2dRdSdTdUdVdWdXeZejdYejZ ejdZejZ d[Z ejd\Z ejd]Z ejd^Zd_Zdad`Zy)bN(#aftncnaftnsepaftnsepc annotation atnauthoratndateatnicnatnid atnparentatnrefatntimeatrfend atrfstartauthor backgroundbkmkend bkmkstartblipuidbuptimcategorycolorschememappingcolortblcommentcompanycreatim datafield datastoredefchpdefpapdodoccommdocvar dptxbxtextebcendebcstart factoidnamefaltfchars ffdeftext ffentrymcr ffexitmcrffformat ffhelptextfflffname ffstattextfilefiletblfldinstfldtypefonttblfnamefontembfontfilefooterfooterffooterlfooterrfootnote formfieldftncnftnsepftnsepcg generatorgridtblheaderheaderfheaderlheaderrhlhlfr hlinkbasehllochlsrchsvhtmltaginfokeycodekeywords latentstyleslchars levelnumbers leveltextlfolevellinkvallist listlevellistname listoverridelistoverridetable listpicture liststylename listtablelsdlockedexceptmaccmaccPr mailmergemalnmalnScrmanagermargPrmbarmbarPrmbaseJcmbegChr mborderBox mborderBoxPrmboxmboxPrmchrmcountmctrlPrmdmdegmdegHidemdenmdiffmdPrmemendChrmeqArrmeqArrPrmfmfNamemfPrmfuncmfuncPr mgroupChr mgroupChrPrmgrowmhideBot mhideLeft mhideRightmhideTopmhtmltagmlimmlimlocmlimlow mlimlowPrmlimupp mlimuppPrmmmmaddfieldnamemmath mmathPictmmathPrmmaxdistmmcmmcJc mmconnectstrmmconnectstrdatammcPrmmcs mmdatasourcemmheadersource mmmailsubjectmmodso mmodsofiltermmodsofldmpdatammodsomappedname mmodsonamemmodsorecipdata mmodsosort mmodsosrc mmodsotable mmodsoudl mmodsoudldatammodsouniquetagmmPrmmquerymmrmnarymnaryPrmnoBreakmnummobjDistmoMath moMathPara moMathParaPrmopEmumphantmphantPrmplcHidemposmrmradmradPrmrPrmsepChrmshowmshpmsPremsPrePrmsSubmsSubPrmsSubSup msSubSupPrmsSupmsSupPr mstrikeBLTRmstrikeH mstrikeTLBRmstrikeVmsubmsubHidemsupmsupHidemtranspmtypemvertJcmvfmfmvfmlmvtofmvtolmzeroAsc mzeroDescmzeroWidnesttablepropsnextfile nonesttablesobjaliasobjclassobjdataobjectobjnameobjsectobjtime oldcprops oldpprops oldsprops oldtpropsoleclsidoperatorpanosepassword passwordhashpgppgptblpicproppictpnpnseclvlpntextpntxtapntxtbprintimprivatepropnameprotend protstart protusertblpxeresultrevtblrevtimrsidtblrxeshpshpgrpshpinstshppictshprsltshptxtsnsp staticval stylesheetsubjectsvsvbtctemplate themedatatitletxeudupr userprops wgrffmtfilter windowcaptionwritereservationwritereservhashxexform xmlattrname xmlattrvaluexmlclosexmlnamexmlnstblxmlopencp1252*M mac_romanN mac_japaneseOmac_chinesetradP mac_koreanQ mac_arabicR mac_hebrewS mac_greekT mac_cyrillicUmac_chinesesimpV mac_rumanianW mac_ukrainianXmac_thaiYmac_cecp932cp949cp1361cp936cp950cp1253cp1254cp1258cp1255cp1256cp1257cp1251cp874cp1250cp437cp850) z )parsectpagelinetab emdashu—endashu–emspaceu enspaceu qmspaceu bulletu•lquoteu‘rquoteu’ ldblquoteu“ rdblquoteu”rowcell|nestcell~  {}\­u‑)ryrzr{r|-_zO\\([a-z]{1,32})(-?\d{1,10})?[ ]?|\\'([0-9a-f]{2})|\\([^a-z])|([{}])|[\r\n]+|(.)zP(\{\\field\{\s*\\\*\\fldinst\{.*HYPERLINK\s(\".*\")\}{2}\s*\{.*?\s+(.*?)\}{2,3})cd|vsd|vr|Sg}d}t|}d}d}||kr|s#|jd|rd}|tdz }*|r|jd|r_|tdz }d}||kr9||jr&|||z }|dz }||kr||jr&t|}||z }||dk(rd}|dz }|s|j |||dz }||krdj |S) z Remove all \pict groups with binary data from the RTF text. If no binary-encoded \pict groups are found, return the original text. See issue 58 z\pictz\binrFTr{)len startswithisdigitintappendjoin)rtf_textrinin_pict binary_length length_strs U/home/spazio/.cache/hermes-doc-venv/lib/python3.12/site-packages/striprtf/striprtf.pyremove_pict_groupsrs9x7(#: F A H AGM a%8..x;G X A  ""7A.S\! !e 3 3 5(1+-JFA!e 3 3 5!$J ]"!#Q MM(1+ & Q; a%> 776?z%\\f(\d+).*?\\fcharset(\d+).*?([^;]+);z{[^{}]*\\fonttblz[{}]c2tj|}|syd}tj||j D]D}||j dk(rdndz }|dk(s#||j |j cS||j dS)z Return the ``{\fonttbl ...}`` group of the RTF text, or "" if there is none. The font table regex is expensive to run against a whole document, so it is only applied to this slice. See issue 71. rrrzrN)FONTTABLE_STARTsearchBRACEfinditerendgroupstart)textrdepthbraces rfont_table_grouprs  " "4 (E  Eeiik25 ekkms*2 A:  4 45    rc Tt|}tjtd|}g}i}d}d}d}d}d} d} d} d} d} d}tj t |}|D]:\}}}|j|tjt||d||<<tj|D]}|j\}}}}}}| rO|sM| tj| j!|j|d|ijd|| z } d} |rUd} |d k(r| dz } d }|j#| ||f|d k(s| dz} |r|j%\} }}nd} d }|s| dks| S|r-d} |t&vr|t(vr|}|r| t&|z } |d k(sd }|rd} |t*vrd }n |dk(rd|} t-j.||s|r'|t&vr| t&|z } =|dk(r t|} O|dk(r-|| } Zt|}|dkr|dz }| t3|z } | } |dk(r|}|dk(r|}|dk(rd }|dk(sd }|r| dkDr| dz} |r| s|} | |z } |s| dkDr| dz} |r|r| |z } | S#t0$rd}YwxYw)aUConverts the rtf text to plain text. Parameters ---------- text : str The rtf text encoding : str Input encoding which is ignored if the rtf file contains an explicit codepage directive, as it is typically the case. Defaults to `cp1252` encoding as it the most commonly used. errors : str How to handle encoding errors. Default is "strict", which throws an error. Another option is "ignore" which, as the name says, ignores encoding errors. Returns ------- str the converted rtf text as a python unicode string z\1(\2)NFrrr)namecharsetencodingr)rerrorsrzTr{*ansicpgcputf8ucuifdeffr5r)rresub HYPERLINKS FONTTABLEfindallrstrip charset_mapgetrPATTERNrgroupsbytesfromhexdecoderpop specialchars sectionchars destinationscodecslookup LookupErrorchr)rrrstackr5 default_font current_font ignorablesuppress_outputucskipcurskiphexesoutr in_documentfonttbl_matchesfont_idfcharset font_namematchwordarg_hexcharrtcharcs r rtf_to_textrs>( d #D 66J D EGLLIO FG E C EK ''(8(>?O(7 $9OO%#H x@  !!$'[.3lln+c4ue  5=='.. \J3IJNN / C E G| "  fiAB# 9>6FI F $I5A:z JyG|#<'#/L <--C G|# "u:&MM(+O%|D))S;$GCA1uW 3q6MC$G" " ""&#"& {1  ETME {1 u w[z JU#&%H&s=J J'&J')r&strict)rr frozensetrrrrcompile IGNORECASErrrrrrrrrrrs )) V x                       ! "# $              ?  HVV<  t ( (  8  8  8 ( ( ( X X t  C V!" d#$    /45 : "**VMM  RZZWMM .` BJJ? @ "**01 7!&Kr