j\"UdZddlZddlmZmZmZmZddlmZddl m Z m Z ddl m Z mZmZdaeed <daeed <daeed <d Zeed <dZdZGddeZ d0deeedfdeeedfdeeeedfdeeeeffdZdeedeedeefdZdeedefdZ dedeeeeefdeeeeefdeeeeefdee deed fd!ed"ed#eeeeeeegdfd$ed%ed&edeeeeeeeffd'Z!d(eeeefd)eed*eed+edeed fdeee"ff d,Z#ded)eed*eedee d+ed-ed"ed.e"d#eeeeeeegdfdeee"effd/Z$dS)1ze Code related to text extraction. Some parts are still in _page.py. In doubt, they will stay there. N)AnyCallableOptionalUnion)Font)is_char_neutral is_char_rtl)DictionaryObjectTextStringObjectencode_pdfdocencodingCUSTOM_RTL_MINCUSTOM_RTL_MAXCUSTOM_RTL_SPECIAL_CHARS LAYOUT_NEW_BT_GROUP_SPACE_WIDTHSiceZdZdS)OrientationNotFoundErrorN)__name__ __module__ __qualname__MC:\Projects\DSCons\.venv\Lib\site-packages\pypdf\_text_extraction\__init__.pyrrsDrr_min_maxspecialsreturnc"t|tr,t|cxkr tkrnnt |ndant|t r|at|tr,t|cxkr tkrnnt |ndant|t r|at|t r|an4t|trd d|Dat ttfS)a Change the Right-To-Left and special characters custom parameters. Args: _min: The new minimum value for the range of custom characters that will be written right to left. If set to ``None``, the value will not be changed. If set to a valid integer, it will be converted to its corresponding character. The default value is "", which sets no additional range to be converted. _max: The new maximum value for the range of custom characters that will be written right to left. If set to ``None``, the value will not be changed. If set to a valid integer, it will be converted to its corresponding character. The default value is "", which sets no additional range to be converted. specials: The new list of special characters to be inserted in the current insertion order. If set to ``None``, the current value will not be changed. If set to a string, it will be converted to a list of ASCII codes. The default value is an empty list. Returns: A tuple containing the new values for ``CUSTOM_RTL_MIN``, ``CUSTOM_RTL_MAX``, and ``CUSTOM_RTL_SPECIAL_CHARS``. rc3hK|]-}t|cxkr tknnt|V.dS)N)UNICODE_LOWER_LIMITUNICODE_UPPER_LIMITchr).0chars r z!set_custom_rtl..DsT+ + .AT.`.`.`.`M`.`.`.`.`.`CII.`.`.`.`+ + r) isinstanceintr"r#r$rstrrrlistjoin)rrrs rset_custom_rtlr-s+>$&9T&X&X&X&XEX&X&X&X&X&XT^` D#  $&9T&X&X&X&XEX&X&X&X&X&XT^` D#  (C   #+  Hd # # #%77+ + "*+ + + $ $   >+C CCrmnc |d|dz|d|dzz|d|dz|d|dzz|d|dz|d|dzz|d|dz|d|dzz|d|dz|d|dzz|dz|d|dz|d|dzz|dzgS)Nrrrr)r.r/s rmultr4Js !qt adQqTk! !qt adQqTk! !qt adQqTk! !qt adQqTk! !qt adQqTk!AaD( !qt adQqTk!AaD(  rcZ|ddkrdS|ddkrdS|ddkrdSdS) Nr2gư>rgưr1Zr)r.s rorientr9Us=td{{qte||staxxr 3rtext cmtm_prev cmtm_matrix memo_cmtm font_resource orientations.output font_size visitor_text str_widths spacewidth str_heightc |d} |d} |d}|d}|d}|d}t| | }t||}t|}|d|dz }|d|dz }tj| ddz| ddzz}tj| ddz| ddzz}tj|ddz|ddzz}|} ||vrt|dvr|}|}n|dvr|}|} t |d t | |z||zzkr.||zd d kr||d zz }|||d z||||d }n || | z|zkr||zd d kr|d z }n#t$rYnwxYw|} |} ||| | fS)Nrr1r3rrr2)rr6)r7r8g? r ) r4r9mathsqrtrabsmin Exceptioncopy)r:r;r<r=r>r?r@rArBrCrDrEcm_prevtm_prev cm_matrix tm_matrixmemo_cmmemo_tmm_prevr. orientationdelta_xdelta_y scale_prev_x scale_prev_yscale_y moved_height moved_widths rcrlf_space_checkr__s>lGlGAIAIlGlG '7 # #F Y ""A))KdVAYGdVAYG9WQZ]WQZ]:;;L9WQZ]WQZ]:;;Li ! a)A,/9::GG,&&&&h% $  ! !   |  sSl)BIPWDW%X%XX X X r"d**$+%+ Lt %!  Z*4 D D D$#s** CKD      nnGnnG ' ))sA8F F"!F"operandsrRrSfontc<d}d}t||}t|}||vrt|dkrt|dtr |d}d}nd}t|dtrt |dn|d} tjtrC | jd}nG#t$r| dd}Yn%wxYwd fd| D}||fS)NrFrT surrogatepasscharmapcg|]:}|jvr j|n!t|f;Sr)encodingbytesdecode)r%xras r z%get_text_operands..sFbbbZ[dm););T]1%%tASASAUAUbbbr) r4r9lenr(r*r rfrhrNr,) r`rRrSrar?tis_str_operandsr.rWtts ` rget_text_operandsrosEAO Y ""A))Kl""s8}}q'8'8 hqk3 ' '  A"OOAhqk3//!%hqk222a[  $--- > $-AAAA >>> )_==AAA > GGbbbb_abbb  s7C C65C6 text_operandsrtl_dirc d} |D]} | | |jkr|jn|| z } |j| | } t | dkrwt | tr |r| |zn|| z}|t| ttr|sd}|||||||d}| |z}|rd}|||||||d}|| z}|r| |zn|| z}||| fS)Ngr1TrF) space_char space_widthget_text_width character_mapgetrkr rr rr) r:rRrSr>rarprArqrBwidths raw_characterris rget_display_strrzsTF&55 mt&F&F$""DL_L_`mLnLnn   " "=- @ @ q66Q;; #;<< $+8q4xxqA~~>> "G#/$ T9iPYZZZD4x#G#/$ T9iPYZZZDax '41t88D1HDD &  r)rrN)%__doc__rJtypingrrrr_fontr_utilsr r genericr r r rr*__annotations__rrrr)r"r#rNrr+tupler-floatr4r9r_boolrorzrrrrs  11111111111111111111OOOOOOOOOO "#"""() #)))     y   #%"$,0-D-D S$ -D S$ -DCcD()-D 3S= -D-D-D-D`DKDKDKd5kc>* >*T%[$u+-.>*tE{DK/0>*T%[$u+-. >* ,- >* S/ >* >*>*8S#sC$=t$CDE>*>*>*>* 3T%[$u+ -.>*>*>*>*B" 5../0" E{" E{"   " S/ "  39 " " " " J,! ,!E{,!E{,!,- ,!  ,!  ,!,!,!8S#sC$=t$CDE,! 3e ,!,!,!,!,!,!r