JՁj,;ddlZddlmZddlmZddlmZmZmZm Z m Z m Z ddl m Z ejeZdZde d efd Zde d efd Zde d efd Zde d efd Zded efdZdeded efdZde ded efdZde d efdZddeeded efdZy)N)BytesIO)IO) WHITESPACESWHITESPACES_AS_BYTESBinaryStreamType StreamTypelogger_warningread_non_whitespace) PdfReadErrori streamreturnct|}||jdz }|jdd|dddk(xr|dddk(xs |ddtvS)NrEI)r readseekr)r ei_toks B/tmp/pip-target-jj2dfcxx/lib/python/pypdf/generic/_image_inline.py_check_end_image_markerr1s]  (F fkk!nF KKA "1:  UF1QK3$6$T&1+:TUrct} t||jtx}z}|r|s t d|j d}|dk\r.||d|dzz }|j t| |zdzdn|j d}|dk\ru|j t| |zdz d|jd}|tvr1|j dd|jd}|dz}|tvr1||d|z }n:t|dk(r||z }t d||ddz }|j dd;t|s t d t|S) z` Extract HexEncoded stream from inline image. The stream will be moved onto the EI. Unexpected end of stream.>rNrrrEI stream not found.) bytearrayr r BUFFER_SIZEr findrlenrrbytes)r data_out read_bytes data_bufferedpos_tokpos_eics r extract_inline__ascii_hex_decoder*8s {H +F3V[[Q\E]7]z^ J:; ;$$T* a<  m! 4 4H KK]++g591 = ##E* Q; KK]++f4q8! < AA{" B"KKN! {"  gv. .H  }  "  %H:; ;M#2&& B1 4 #6 *122 ?rct} t||jtx}z}|r|s t d|j d}|dk\r.||d|dzz }|j t| |zdzdn9t|dk(r||z }t d||ddz }|j ddt|s t dt|S) zY Extract A85 stream from inline image. The stream will be moved onto the EI. rs~>rNrrrr) rr rr r r!rr"rr#)r r$r%r&r's rextract_inline__ascii85_decoder,^s {H +F3V[[Q\E]7]z^ J:; ;$$U+ a<  m! 4 4H KK]++g591 =  }  "  %H:; ;M R    B " #6 *122 ?rc0t} |jt}|s td|j d}|dk\r||dz|dz}|j ds|j dr.||d|dzz }|jt| |zdzdnQtdt |j d}|dkDr*||d|z }|jt| |zdz dn||z }t|s td t|S) zj Extract RL (RunLengthDecode) stream from inline image. The stream will be moved onto the EI. rrrrNz=Early EOD in RunLengthDecode of inline image, using fallback.)sourcer) rrr r r! startswithendswithrr"r __name__rr#)r r$r&r' after_token ei_markers r!extract_inline__run_length_decoder6{s. {H  K0 :; ;$$W- a<(! gkBK%%e, 0D0DU0KM-GaK88 S//'9A=qA^gop)..u5 q= k ::HKK]!3 3i ?! CQG M!3 6 #6 *122 ?rczdtdtffd }t}d} |d}|s|dk(r||z }|dk7rd}|d}||z }|dk(rjddn9|d k(rn3|d k(rn.|d vr)|d }||z }|d dz|dz}|||d z z }{t s t dt|S)z` Extract DCT (JPEG) stream from inline image. The stream will be moved onto the EI. lengthrc`j|}|t||k7r td|S)Nr)rr"r )r8_resultr s rrz(extract_inline__dct_decode..reads4++f% ?c'lf4:; ;rFTrs&rrr)intr#rrrr )r rr$ not_firstr)szs` rextract_inline__dct_decoderCs SU{HI  G g MH <  GA  < KKA  '\  '\   T QA MH1ad"B R!V $H/ 2 #6 *122 ?r stream_outpositionc|dkry|j}|j|dz |jdtv}|j||S)a3 Check whether the byte in front of `position` is a whitespace. Returns: Whether the preceding byte is whitespace. If no preceding byte exists, returns ``False`` because the ``EI`` marker is at the beginning of the image data and is therefore not delimited by anything. rFr)tellrrr)rDrEsaved_position is_whitespaces r_is_preceded_by_whitespacerJsQ1}__&NOOHqL!OOA&+5MOON# r first_bytec|j}|dk(r!|jd}|dk(xs|tv}n8|dk(r1|jd}|dddk(xr|dddk(xs |ddtv}nd }|j|d |S) a< Check whether `first_byte` starts an exact `Q` or `EMC` operator. `first_byte` has already been consumed from the stream, which is restored to its current position afterwards. Looking at this byte alone is not sufficient, as binary image data can easily be mistaken for one of these operators. QrrErNrsMCFr)rGrrr)r rKrEsuffixis_closing_operators r _is_followed_by_closing_operatorrQs{{}HTQ$mDv/D t Q$RajE1hva{c7I7gVTUVW[\gMg# KK! rct} |jt}|s td|j d}|dk(r|j |n|j |d|dz|j dz }|j|dzt|z d|j }|jd}|dk7r|j|d|jd}|dk(r7|j|dz d|j| |jS|tvr|j|d9|tvr|jd}|tvrt||s t||s|j|dt|r|j|d|j|dz d|j| |jS)zLegacy method, used by defaultrrNr<rrIr)rrr r r!writerGrr"truncaterrJrQis_followed_by_binary_datagetvalue)r rDr&r( sav_pos_ei saved_postok2tok3s rextract_inline_defaultr\sJ  K0 :; ;##   R<   ] +   ]1vz: ;#*Q.J KK S%77 ; I;;q>Dt| Iq);;q>Ds{ IM1-##J/2     1;& Iq)+%{{1~+% .j*ENnO Iq))&1 Iq) KK A q )    +      m rr8c|j}|j|}|j||syd}d}t|D]+\}}|dkr |tvry|tv}||s|}|&|s)|}n|y||}||z } |||} | j dr| dkDry| j ddjry| d kDryy) a Check if the next bytes of the stream look like binary image data or regular page content. This is just some heuristics due to the PDF specification being too imprecise about inline images containing the `EI` marker which would end an image. Starting with PDF 2.0, we finally get a mandatory length field, but with (proper) PDF 2.0 support being very limited everywhere, we should not expect to be able to remove such hacks in the near future - especially considering legacy documents as well. The actual implementation draws some inspiration from https://github.com/itext/itext-java/blob/9.1.0/kernel/src/main/java/com/itextpdf/kernel/pdf/canvas/parser/util/InlineImageParsingUtils.java FN T/r.rr)rGrr enumeraterr1replaceisdigit) r r8rEdataoperator_start operator_endindexbyterIoperator_lengthoperators rrVrV7s{{}H ;;v D KK NL   t "9%99 44  !-"N  %-!L   "^3ON<0H4 _q%8c"**, r) )loggingiortypingr_utilsrrrr r r errorsr getLoggerr3loggerr boolrr#r*r,r6rCr@rJrQr\rVrrrus8"   8 $ VJV4V#Z#E#L:%:$j$U$N,'7,E,^7cd(ZUt*:!::!%:!z6r%y6#6t6r