mj\tUdZddlZddlmZddlmZmZmZmZm Z ddl m Z m Z ddl mZmZmZddl mZdd lmZmZmZdd lmZmZmZmZmZmZej<ddd k\rdd lmZndd l mZ dd l!m"Z"m#Z#edZ%ee&d<dZ'e(e&d< d,dee)e*eefde(de%de(de+e%e,ff dZ-de.de+e(e(fde(de.fdZ/de)de+e(e(fde.de"jDfdZ0dede+eeeeffdZ1de+e(e(fde.de%dee)efd e(d!e)de+e"jDe)e)e,ffd"Z2de+e(e(fde.de%de)d e(de+e"jDe)e)e,ff d#Z3d$e"jDd%e4e)efd&edee)e*eefd'e,de"jDf d(Z5d%e4e)efd e(dee)e*eefde+e%e,ffd)Z6 d-d%e4e)efd*ee4e)efdfde+ee)e.effd+Z7y#e$$r e$dwxYw).z1Functions to convert an image XObject to an imageN)BytesIO)AnyLiteralOptionalUnioncast)check_if_whitespace_onlylogger_warning) ColorSpacesImageAttributesStreamAttributes) FilterTypes)EmptyImageDataErrorLimitReachedError PdfReadError) ArrayObjectDecodedStreamObjectEncodedStreamObject NullObjectTextStringObjectis_null_or_none) ) TypeAlias)ImageUnidentifiedImageErrorz]pillow is required to do image extraction. It can be installed via 'pip install pypdf[image]') 1RGB2bits4bitsPLRGBACMYK mode_str_typerMAX_IMAGE_MODE_NESTING_DEPTH color_spacecolor_components prev_modedepthreturnc|tkDr tdt|ryd}t|tr|}n^t|t s td||sy|dj dr d|ddd z}n|dd k(r9|d j}tt|d }|jd d}n|ddk(r1|d j}t||||d z\}}|dvrd}||fS|ddk(r+|dj}t||||d z\}}|dfS|ddk(ri|}t|d }|dj}|dk(r)|d k(r$|d ddk7rtdt|d dyt||||d z\}}||fSdddddddd } t | j} | j|xs"d|cxkrt| krnn| |nd xs|}||dk(fS)!z Returns: Image mode, not taking into account mask (transparency). ColorInversion is required (like for some DeviceCMYK). z^Color spaces nested too deeply. If required, consider increasing MAX_IMAGE_MODE_NESTING_DEPTH.)rFrzCannot interpret color spacerz/Calz/DeviceN /ICCBasedz/Nz /Alternate/Indexed)r r&r# /Separationr Tz/DeviceN /DeviceCMYKz/BlackzGColor %(color)s converted to Gray. Please share PDF with pypdf dev team)sourcecolor)r$Trr$r r&r!r")1bitz /DeviceGraypalettez /DeviceRGBr42bit4bit)r(rr isinstancestrlist startswith get_objectrintget_get_image_modelenr __name__values) r)r*r+r,color_space_str icc_profilemode invert_colororiginal_color_spacemode_map mode_valuess O/root/.hermes/venv/lib/python3.12/site-packages/pypdf/generic/_image_xobject.pyrBrB)s ++ l  {#O+s#%  T * *K   Q " "6 *#k!nQR&88 Q; &!!n//1 [%67%//,; Q: %%a.335, -y%!) l ? "D\!! Q= (%a.335, -y%!) lTz Q: %*{1~.%a.335 m +0@A0E#A&q)X5]#.q1!4 , -y%!) l\!!*Hx()K _% $7s;'77 ( )      datasizebitscddlm}|d|dz}||kDrtd|d|dt|}d|zdz }|d|d|zdzdzz}t |x}|krt d t |d ||z zz }d} d|z } t|dD]S} | d|z k7r | dz } d|z } t|dD].} || | z |z|| | |dzz<| |z} | dks%| dz } d|z } 0Ut|S) NrFLATE_MAX_BUFFER_SIZEr1zRequested buffer size z exceeds limit of .z.Image data is not rectangular. Adding padding.)r5) pypdf.filtersrTr bytearrayrCr rDrangebytes) rOrPrQrT buffer_size byte_buffermaskrequiredlength data_indexbityxs rM bits2byterfsT3q'DG#K**"8 EWXmWnno pqqK(K I?DAw47T>A-!34Hd)x'GPXY 8f,--J d(C 47^  !d(? !OJd(CtAw A,0,<,Ct+KKAQK ( 4KCQwa $h     rNrHc |d|dz}t|}||z}ddlm}||kDrtd|d|d t j |||}|S#t $rf}t|} | dk(r td|| |zdk7r|| z dj fd |D}t j |||}Yd}~|Sd}~wwxYw) Nrr1rSzRequested image buffer size z exceeds limit rUz9Data is 0 bytes, cannot process an image from empty data.rNc3NK|]}t|ftzywN)r\r@).0reks rM z$_image_from_bytes..s ;qdSVm,;s"%) rCrYrTrr frombytes ValueErrorrjoin) rHrPrO pixel_countbytes_per_pixelrequired_byte_countrTimgexc data_lengthrks @rM_image_from_bytesrvsq'DG#K$iO%7322*+>*?OdNeef g   0oodD$/ J  0$i ! %K   $ )  + -xx;d;;oodD$/ J 0sA CAB>>Ccbt|}|dk(rd|D\}}}}||||fS|d}t|tr|n|j}|dk(rOd|vrK|dj|dj}}|j d\}}t |}||||fSt d|d |) Nr/c3<K|]}|jywri)r?)rjvalues rMrlz*__handle_flate__indexed..s+X5E,<,<,>+Xsr1rrr z(Expected color space with 4 values, got z: )rCr;r<r?splitr@r)r)countbasehivallookupelement1s rM__handle_flate__indexedrs  E z+XK+X( T5&D%//1~H%h4x(:M:M:OH zf()!n779;q>;T;T;VV nnV, eE D%// A%;-X YYrNcolors obj_as_textc d}d}t|tr|ddk(rt|\}}} |dk(rd}t||d}n|dk(rd}t||d }t |||} |dk(rtt t frjttr jttrj d d d d d dtddd\} } }| jdk(r| dk(rtdt|n2| jdk(rd| z} t!}|| k7rT|| kr)tdt| |t#dg| |z zz n!t%| dstdt| |d| d| | dg}| j'd}t)|d|dDcgc]!\}}|j+|gdz|gdzz#}}}| dk(r|dnt-j.||} n?| j'| } t! dz| zk7rtdt|dn|dk(r.dj1Dcgc]}t#|||gc}d}n|dk(rg}fdt3dd t!d zzd DD]{\}}}}t5d d|d z z zd|d z z z}t5d d|d z z zd|d z z z}t5d d|d z z zd|d z z z}|j7t#|||f}dj1|d}| j9|!| j'|t:j<k(rdnd} nDtA|s9|dd#k(r1|dk7r,t|||d}||k7rt-jB|||} |dk(rd$}d%}| ||d&fScc}}wcc}w#t>$rtd"tdY9wxYw)'zh Process image encoded in flateEncode Returns img, image_format, extension, color inversion PNG.pngrr2r!r#r r"r/)rrr)r1r#r$)rr#r )r/r#r&)rr$r#r r&rrz`Cannot apply lookup for base %(base)s to image with mode 1. Please share PDF with pypdf dev team)r5r}zLNot enough lookup values: Expected %(expected_count)d, got %(actual_count)d.)r5expected_count actual_countNzJToo many lookup values: Expected %(expected_count)d, got %(actual_count)d.r$r1z'Invalid Lookup Table in %(obj_as_text)sr5rrNr r&c3.K|] }||dzyw)r/N)rjnrs rMrlz _handle_flate..3s!+./q1q5)+s)rawmodez=Base %(base)s not coded. Please share PDF with pypdf dev teamr0.tifTIFFF)"r;rrrfrvrrget_dataroriginal_bytesr<encoderBrHr rDrCr\r convertzippointrmergeror[r@append putpaletter DEVICE_GRAYKeyErrorrrm)rPrOrHr)rr image_format extensionr}r~rsnbconvrr colors_arrr5 low_value high_valuebandsb_rgb_c_m_y_k_r_g_bmode2rs @rM _handle_flatersmLI+{+ A*0L+B;+O( T5& wtQ' tQ' D$ -Cj f24GH I__&F f. /**F fc "]]_FO Q " &(  dAr*1- /NBdxx327;#  S!"R"6{ >1#n4&j#++9)5  %~ /L(M"NN5f^_6MN&h#++9)5  $O^4F$Sbk6"#;7 S) 25Z]JqM1R . :LL" c)ZL3,>> #%'eAhu{{4/Gkk$'v;519"22"#LU]kvw!FS[ XX&HAuaAY'7&HIF DV^D+38AVPQAQqAEu}oodD$7 v~   i ..['Im  O  F  s O :&N>O O)(O)ctjt|d}t|||\}}|dk(rt t |j }|dk(}|j dk(r|dk(rd}|j |k(s|j |hddhk(r|}nT|j |hddhk(r0tj||j|j}n|j|}|j dk(r|d k(r|jd}|d d |fS) zc Process image encoded as JPX/JPEG2000 Returns img, image_format, extension, inversion )JPEG2000formatsrr&r%r r$r#r0r.jp2) ropenrrBrr'rHrmrPtobytesr)rPrOrHr)rimg1rIrss rM _handle_jpxrOs 74=-HD(fdCD, rzM499-v~  yyFtu} yyDTYY-#s; ))T vv. .oodDIIt||~>ll4  xx6k[8kk%   FL 00rNrsx_objectlfiltersrIcx tj|vr|tjn|jdk(r:|tjk(r'|sddgnddgt |j znT|jdk(r|tjk(s|r.|jdk(rddgt |j zndt|tr|djdk(rdt|tr4|djdk(rddgt |j ztfdtt Ds_g}tdt d D]3}| |d z|j fd td D5|j|}|S) Nr&g?gr$rr2r3c34K|]}||dzk(yw)r Nr)rjidecodes rMrlz _apply_decode..s%UQfQi1q5&8%Usr r1c3RK|]}td|dz z zzz yw)go@N)round)rjjdmaxdmins rMrlz _apply_decode..s2FGeq5yD4K84?@As$')r DECODErHFT JPX_DECODErCgetbands DCT_DECODEr;rr?allr[extendr) rsrrr)rIlutrrrrs @@@rM _apply_decoderrs)/001 V BMM 9#/#sc3Z#clln-- ((f R]]!: QTQYQY]`Q`sc#,,.11 ; , N % % ': 5; , N % % '= 8sc#,,.11 #%U%F BT%U"Uq#f+q) A!9D!a%=D JJKPQT:   iin JrNc>tj|vr&|tjtjk(rd}|j dddkr't |j ddddd\}}||fSt ||dk(rt |sd|vrd n|d\}}||fS) Nr z/BitsPerComponentrWrcrrr1Grayr )r COLOR_SPACEr DEVICE_RGBrArBr)rrr)rHrIs rM_get_mode_and_invert_colorrs ##x/ _00 1[5K5K K$||'+a/,||/34C 8!R l"  - ! ' 4k1    l  rNpillow_parametersc dtjdtttfdtdtdtdttjttff d}||j j n|j }tt|tjtt|tjf}|j}t|tr|j}t||d |d zzd k(r |d d k(r|dd }|j!d d }|j!dt#j%}t|t&r!t|d k(r|d j%}t)|||\}} |j!t*j,t#j%} t| t&r| d n| } |j!t*j.} | rt| tt&fr| d } ni} t| tsi} d} | t0j2t0j4fvrt7||||||\}}} }n~| t0j8t0j:fvr?| t0j8k(rd}d} nd}d} tj<t?|d}n| t0jDk(r#tj<t?|dd} }}n| t0jFk(rtI|||||\}}} } n| t0jJk(r*tj<t?|ddddf\}}} } n| t0jLk(r*tj<t?|ddddf\}}} } nE|dk(rtC|||dddf\}}} } n)|dk(rtOd|tC|||dddf\}}} } tQ||| || }|||||| \}} }|i}|dk(rd|_)d|vrd|d<t?} |jT|fd |i||j[} tj<t?|}| ||fS#t@$rtC|||}YwxYw#tV$r5|jYd!}d}d} t?}|jU||"YwxYw#t\$r}t_d#t`|$d}Yd}~d}~wwxYw)%a Users need to have the pillow package installed. It's unclear if pypdf will keep this function here, hence it's private. It might get removed at any point. Args: x_object: pillow_parameters: parameters provided to Pillow Image.save() method, cf. Returns: Tuple[file extension, bytes, PIL.Image.Image] rsrrrrr-ctj|vrt|tjd}|j|jk7rt dt |nr|j dk7r|jd}|j dk(r|jd}n |j dk(r|jd}|j|d|vrd }d }nd }d }|||fS) Nr z1image and mask size not matching: %(obj_as_text)srr$r#r rJPEGrrrr) r S_MASK_xobj_to_imagerPr rDrHrputalpha)rsrrrralphas rM _apply_alphaz$_xobj_to_image.._apply_alphas  ! !X -"8O,B,B#CDQGExx5::%G# +::$!MM#.E88s?++e,CXX_++c*C U#%) " $ " I|++rNNrr1rz/Colorsz /ColorSpacerz.tiffrr)rrrrz.jpg)rF)rPPMr&rrzColorSpace field not found in qualitykeepformatr%)rz#Failed loading image: %(exception)s)r5 exception)1rdictr<rtupleindirect_reference__repr__rr@r WIDTHHEIGHTrr;rrCrArr?r=rrFILTER DECODE_PARMSr FLATE_DECODERUN_LENGTH_DECODEr LZW_DECODEASCII_85_DECODErrrrvrrrCCITT_FAX_DECODE JBIG2_DECODErrrsaveOSErrorrgetvalue Exceptionr rD)rrrrrPrOrr)rHrIfiltersr decode_parmsrrsr_ img_byte_arrrrs rMrrs&, [[,sCx.,, ,  , u{{C$ % ,F   ##,,.     h445 6S(?KaKaBb8c dD    D${{} 4yDGd1g%&!+RD0@CRy\\)Q 'F||M:<@KKMK+t$[)9Q)>!!n//1 3HfkRD,ll+22JLALLNG($7wr{WH<< 0 = =>L <%?#A  lD ) IBOOR%9%9::*7       + '\9a bmmR%7%78 8 r}} $!LI LI 6**WT]ODC R]] "',zz'$-'@&&9\ R]] "5@ $k66 2\9l R(( ( JJwt}i 8    6 2\9l R__ $ JJwt}n =    6 2\9l  dD$ /    6 2\9l ;H:FGG dD$ /    6 2\9l Xxl KC#/ X{L)$ CL v - -+1 i (9L5HlH6GH  "Djj' dC S& 6#D$5C 6v 5{{6"  y  ,| 4 5 >T)rri)8__doc__sysiortypingrrrrr_utilsr r constantsr r rrrerrorsrrrgenericrrrrrr version_infortyping_extensionsPILrr ImportErrorr'__annotations__r(r@r<r=rboolrBr\rfrvrrrrrrrrrNrMrs77 66=FF)IIBQ7" +1#> y%'c& U sDIs*+U U U   U  =$  U pEsCx< 38_,1 [[:ZZsCc?Q9RZ$~/ S/~/ ~/ ~/sK'( ~/  ~/  ~/ 5;;S$ &'~/B 1 S/ 1  1  1 1  1  5;;S$ &' 1F( (38n((sDIs*+ (  (  [[ (V38n&)8=c49c>Q8R =$ @:>t sCx.t  c3h!56t  8C=% $%t    = s .F))F7