o h@s4ddlZddlZddlmZddlmZmZmZmZm Z m Z m Z m Z ddl mZddlmZmZmZmZmZddlmZmZmZddlmZmZddlmZmZdd lm Z m!Z!dd l"m#Z#m$Z$m%Z%m&Z&m'Z'm(Z(dd l)m*Z*dd l+m,Z,m-Z-m.Z.m/Z/m0Z0m1Z1m2Z2dd l3m4Z4m5Z5ddl6m7Z7m8Z8m9Z9m:Z:m;Z;mZ>ddl?m@Z@mAZAmBZBmCZCmDZDmEZEmFZFeGeHZIGddde ZJGddde ZKe8dZLe8dZMe8dZNe8dZOe8dZPGdddZQe eRe eReReRfe eReReReRffZSGdddZTGdddZUGdd d e;e e9e.fZVe) rMrNrOrPrQrRrSrTrIrJrVr?r?r@__repr__VszPDFTextState.__repr__cCsZt}|j|_|j|_|j|_|j|_|j|_|j|_|j|_|j|_|j |_ |j |_ |SN) rHrMrNrOrPrQrRrSrTrIrJrWobjr?r?r@copyiszPDFTextState.copycCst|_d|_dS)Nrr)r2rIrJrVr?r?r@rUw zPDFTextState.resetrKN)rKrH) r<r=r>r3__annotations__r5rXstrrYr]rUr?r?r?r@rHEs   rHc@s.eZdZd ddZd ddZdefddZdS) PDFGraphicStaterKNcCsNd|_d|_d|_d|_d|_d|_d|_d|_td|_ d|_ td|_ dS)Nr DeviceGray) linewidthlinecaplinejoin miterlimitdashintentflatnessscolorrscsncolorncsrVr?r?r@rXs zPDFGraphicState.__init__cCsRt}|j|_|j|_|j|_|j|_|j|_|j|_|j|_|j|_|j |_ |SrZ) rcrerfrgrhrirjrkrlrnr[r?r?r@r]szPDFGraphicState.copyc Cs,d|j|j|j|j|j|j|j|j|jf S)Nz) rerfrgrhrirjrkrlrnrVr?r?r@rYszPDFGraphicState.__repr__r`)rKrc)r<r=r>rXr]rbrYr?r?r?r@rcs   rcc@speZdZdZddeddfddZdeeddfd d Zdd e d ede fddZ dede e efde fddZdS)PDFResourceManagerzRepository of shared resources. ResourceManager facilitates reuse of shared resources such as fonts and images so that large objects are not allocated multiple times. TcachingrKNcCs||_i|_dSrZ)rq _cached_fonts)rWrqr?r?r@rXr_zPDFResourceManager.__init__procscCs"|D] }|tus |tur q qdSrZ) LITERAL_PDF LITERAL_TEXT)rWrsprocr?r?r@ get_procsets zPDFResourceManager.get_procsetFcmapnamestrictcCs.zt|WStjy|rtYSwrZ)rget_cmap CMapNotFoundr)rWrxryr?r?r@rzs  zPDFResourceManager.get_cmapobjidspeccCsL|r||jvr|j|}|Std||tjr"|dtur"tdd|vr-t|d}n tjr4tdd}|dvr@t||}nZ|dkrJt ||}nP|d krTt ||}nF|d vr^t ||}n<|d krt |d }|slJt |d }dD]}||vrt||||<qv|d|}ntjrtd|t||}|r|jr||j|<|S)Nz#get_font: create: objid=%r, spec=%rTypezType is not /FontSubtypezFont Subtype is not specified.Type1)rMMType1TrueTypeType3) CIDFontType0 CIDFontType2Type0DescendantFontsr)Encoding ToUnicodezInvalid Font spec: %r)rrlogdebugr STRICT LITERAL_FONTrr1rrrrr%r$r]r&get_fontrq)rWr|r}rMsubtypedfontssubspeckr?r?r@rsF (          zPDFResourceManager.get_font)T)F)r<r=r>__doc__boolrXrobjectrwrbrrzrrrr?r?r?r@rps "rpc @seZdZdeeddfddZdddZdeddfd d Zdd d Z ddede de ee ffddZ dddZ edZedZed ZdededdfddZdS)PDFContentParserstreamsrKNcCs||_d|_t|ddSNr)ristreamr.rX)rWrr?r?r@rXszPDFContentParser.__init__cCsT|js(|jt|jkrt|j|j}|jd7_ntdt||_dSdS)NzUnexpected EOF, file truncated?)fprlenrr'r(rget_data)rWstrmr?r?r@fillfpszPDFContentParser.fillfpposcCs|t||dSrZ)rr.seek)rWrr?r?r@rszPDFContentParser.seekcCsT|jt|jkr dS ||j|_|j|j|_|jr!nd|_q d|_dS)Nrr) charposrbufrrtellbufposreadBUFSIZrVr?r?r@fillbufs  zPDFContentParser.fillbufEItargetcCsB||d}d}|t|kr||rL|j|j}t|f}||7}|jd7_t||kr5|sD|t|krI|t||fkrI|d7}nQsz/PDFContentParser.do_keyword..rFs~>sID )r) KEYWORD_BI start_type KEYWORD_IDend_typerr)r7get isinstancer-r!rr#push KEYWORD_EIr r) rWrr_objs error_msgdeosfilterrr\r?r?r@ do_keywordGs<        zPDFContentParser.do_keywordr`)r)r<r=r>rrrXrintrrrrrrr*rrrr,rr?r?r?r@rs   !rc@seZdZdZdededdfddZddd Zd ee e fddfd d Z d e ddfddZ de ddfddZdedee fddZdee eeffddZdee eefddfddZdddZdddZde d e d!e d"e d#e d$e ddfd%d&Zd'e ddfd(d)Zd*e ddfd+d,Zd-e ddfd.d/Zd0e ddfd1d2Zd3e d4e ddfd5d6Zd7e ddfd8d9Zd:e ddfd;d<Z d=e ddfd>d?Z!d@e dAe ddfdBdCZ"d@e dAe ddfdDdEZ#dFe dGe dHe dIe dJe dKe ddfdLdMZ$dHe dIe dJe dKe ddf dNdOZ%dFe dGe dJe dKe ddf dPdQZ&ddRdSZ'd@e dAe dTe dUe ddf dVdWZ(ddXdYZ)ddZd[Z*dd\d]Z+dd^d_Z,dd`daZ-ddbdcZ.ddddeZ/ddfdgZ0ddhdiZ1ddjdkZ2ddldmZ3ddndoZ4d=e ddfdpdqZ5d=e ddfdrdsZ6dte ddfdudvZ7dte ddfdwdxZ8dye dze d{e ddfd|d}Z9dye dze d{e ddfd~dZ:de de dAe de ddf ddZ;de de dAe de ddf ddZdddZ?dddZ@d=e ddfddZAdddZBdddZCdddZDdddZEde ddfddZFde de ddfddZGde ddfddZHde de ddfddZIdddZJde ddfddZKde ddfddZLde ddfddZMde ddfddZNde de ddfddZOde ddfddZPde ddfddZQde de ddfddZRde de ddfddZSde d{e de de de de ddfddńZTdddDŽZUde ddfddʄZVde ddfdd̈́ZWde ddfddτZXde de de ddfddӄZYdddՄZZdddׄZ[de ddfddلZ\de ddfdd܄Z]de^ddfdd߄Z_e`fd ee e fdeae d e ddfddZbdeae ddfddZcdS)PDFPageInterpreterzhProcessor for the content of a PDF page Reference: PDF Reference, Appendix A, Operator Summary rsrcmgrdevicerKNcCs||_||_dSrZ)rr)rWrrr?r?r@rXqr_zPDFPageInterpreter.__init__cCs||j|jSrZ) __class__rrrVr?r?r@dupuzPDFPageInterpreter.dup resourcesc Cs2||_i|_i|_t|_|sdSdtdttfdd}t | D]s\}}t d|||dkrVt | D]\}}d}t |trF|j}t |}|j|||j|<q8q#|dkrut | D]\}}|t|} | durs| |j|<q`q#|d kr|jt|q#|d krt | D] \} } | |j| <qq#dS) z@Prepare the fonts and XObjects listed in the Resource attribute.Nr}rKcSst|tr t|d}nt|}|dkr*t|tr*t|dkr*t|t|ddS|dkrDt|trDt|dkrDt|tt|dSt|S)NrICCBasedrrNDeviceN) rlistr1rrr'r%rr)r}namer?r?r@get_colorspaces  z9PDFPageInterpreter.init_resources..get_colorspacezResource: %r: %rrE ColorSpaceProcSetXObject)rfontmapxobjmaprr]csmaprrrr$itemsrrrr"r|rrr&rwr%) rWrrrrfontidr}r|csid colorspacexobjidxobjstrmr?r?r@init_resourcesxs>      z!PDFPageInterpreter.init_resourcesctmcCs:g|_||_|j|jt|_t|_g|_g|_ dS)zrVr?r?r@do_f~rAzPDFPageInterpreter.do_fcCr)z6Fill path using nonzero winding number rule (obsolete)Nr?rVr?r?r@do_FrzPDFPageInterpreter.do_FcCr=)zFill path using even-odd ruleFTNr>rVr?r?r@do_f_arAzPDFPageInterpreter.do_f_acCs"|j|jddd|jg|_dS)z6Fill and stroke path using nonzero winding number ruleTFNr>rVr?r?r@do_BrAzPDFPageInterpreter.do_BcCs"|j|jddd|jg|_dS)z(Fill and stroke path using even-odd ruleTNr>rVr?r?r@do_B_arAzPDFPageInterpreter.do_B_acCrB)z>Close, fill, and stroke path using nonzero winding number ruleN)r7rHrVr?r?r@do_brDzPDFPageInterpreter.do_bcCrB)z0Close, fill, and stroke path using even-odd ruleN)r7rIrVr?r?r@do_b_arDzPDFPageInterpreter.do_b_acCs g|_dS)z$End path without filling or strokingN)rrVr?r?r@do_ns zPDFPageInterpreter.do_ncCr)z3Set clipping path using nonzero winding number ruleNr?rVr?r?r@do_WrzPDFPageInterpreter.do_WcCr)z%Set clipping path using even-odd ruleNr?rVr?r?r@do_W_arzPDFPageInterpreter.do_W_acC@z |jt||j_WdStytjrtd|YdSw)zOSet color space for stroking operations Introduced in PDF 1.1 Undefined ColorSpace: %rN)rr1rrmKeyErrorr rrBrr?r?r@do_CSs  zPDFPageInterpreter.do_CScCrO)z*Set color space for nonstroking operationsrPN)rr1rrorQr rrBrr?r?r@do_css  zPDFPageInterpreter.do_csgraycC@t|}|durtd|ddS||j_|jd|j_dS)z&Set gray level for stroking operationsNCannot set gray level because rrd)r rrrrlrrmrWrTgray_fr?r?r@do_G zPDFPageInterpreter.do_GcCrU)z)Set gray level for nonstroking operationsNrVrrd)r rrrrnrrorWr?r?r@do_grZzPDFPageInterpreter.do_grgbcCJt|||}|durtd|||fddS||j_|jd|j_dS)z%Set RGB color for stroking operationsNz6Cannot set RGB stroke color because not all values in r DeviceRGB)rrrrrlrrmrWr\r]r^rgbr?r?r@do_RG zPDFPageInterpreter.do_RGcCr_)z(Set RGB color for nonstroking operationsNz:Cannot set RGB non-stroke color because not all values in rr`)rrrrrnrrorar?r?r@do_rgrdzPDFPageInterpreter.do_rgrrrcCNt||||}|durtd||||fddS||j_|jd|j_dS)z&Set CMYK color for stroking operationsNz7Cannot set CMYK stroke color because not all values in r DeviceCMYK)r rrrrlrrmrWrrrrcmykr?r?r@do_KzPDFPageInterpreter.do_KcCrf)z)Set CMYK color for nonstroking operationsNz;Cannot set CMYK non-stroke color because not all values in rrg)r rrrrnrrorhr?r?r@do_krkzPDFPageInterpreter.do_kcC|jjj}||}t||krtd|d|ddSt|dkrA|d}t|}|dur;td|ddS||j_dSt|d kr`t |}|durZtd |d dS||j_dSt|d krt |}|durytd |ddS||j_dStdt|ddS)z"Set color for stroking operations.z)Cannot set stroke color because expected  components but got !rrrNz%Cannot set gray stroke color because rz/Cannot set RGB stroke color because components  cannot be parsed as RGBz0Cannot set CMYK stroke color because components  cannot be parsed as CMYKz Cannot set stroke color because T components are specified but only 1 (grayscale), 3 (rgb) and 4 (cmyk) are supported) rrm ncomponentsrrrrr rlrr rWr componentsrTrXrbrir?r?r@do_SCN>            zPDFPageInterpreter.do_SCNcCrm)$Set color for nonstroking operationsz-Cannot set non-stroke color because expected rnrorrNz)Cannot set gray non-stroke color because rrpz3Cannot set RGB non-stroke color because components rqrrz4Cannot set CMYK non-stroke color because components rsz$Cannot set non-stroke color because rt) rrorurrrrr rnrr rvr?r?r@do_scn1ryzPDFPageInterpreter.do_scncC |dS)z!Set color for stroking operationsN)rxrVr?r?r@do_SC^r zPDFPageInterpreter.do_SCcCr|)rzN)r{rVr?r?r@do_scbr zPDFPageInterpreter.do_sccCr)z%Paint area defined by shading patternNr?rr?r?r@do_shfrzPDFPageInterpreter.do_shcC|jdS)zBegin text object Initializing the text matrix, Tm, and the text line matrix, Tlm, to the identity matrix. Text objects cannot be nested; a second BT cannot appear before an ET. N)rrUrVr?r?r@do_BTiszPDFPageInterpreter.do_BTcCr)zEnd a text objectNr?rVr?r?r@do_ETrrzPDFPageInterpreter.do_ETcCr)zBegin compatibility sectionNr?rVr?r?r@do_BXurzPDFPageInterpreter.do_BXcCr)zEnd compatibility sectionNr?rVr?r?r@do_EXxrzPDFPageInterpreter.do_EXtagcC0t|tr |j|dStd|ddS)zDefine marked-content pointz+Cannot define marked-content point because  is not a PSLiteralNrr-rdo_tagrrrWrr?r?r@do_MP{  zPDFPageInterpreter.do_MPpropscC2t|tr|j||dStd|ddS)z.Define marked-content point with property listz>Cannot define marked-content point with property list because rNrrWrrr?r?r@do_DP  zPDFPageInterpreter.do_DPcCr)zBegin marked-content sequencez-Cannot begin marked-content sequence because rNrr-r begin_tagrrrr?r?r@do_BMCrzPDFPageInterpreter.do_BMCcCr)z0Begin marked-content sequence with property listz@Cannot begin marked-content sequence with property list because rNrrr?r?r@do_BDCrzPDFPageInterpreter.do_BDCcCr)zEnd marked-content sequenceN)rend_tagrVr?r?r@do_EMCszPDFPageInterpreter.do_EMCspacecCr)zSet character spacing. Character spacing is used by the Tj, TJ, and ' operators. :param space: a number expressed in unscaled text space units. Nz(Could not set character spacing because r)r rrrrO)rWrrOr?r?r@do_Tc   zPDFPageInterpreter.do_TccCr)zSet the word spacing. Word spacing is used by the Tj, TJ, and ' operators. :param space: a number expressed in unscaled text space units Nz#Could not set word spacing becuase r)r rrrrP)rWrrPr?r?r@do_TwrzPDFPageInterpreter.do_TwscalecCr)zuSet the horizontal scaling. :param scale: is a number specifying the percentage of the normal width Nz)Could not set horizontal scaling because r)r rrrrQ)rWrscale_fr?r?r@do_Tz   zPDFPageInterpreter.do_TzrRcCs4t|}|durtd|ddS| |j_dS)zSet the text leading. Text leading is used only by the T*, ', and " operators. :param leading: a number expressed in unscaled text space units Nz#Could not set text leading because r)r rrrrR)rWrR leading_fr?r?r@do_TLs  zPDFPageInterpreter.do_TLrrNcCsz |jt||j_Wnty&tjrtd||j di|j_Ynwt |}|dur:t d|ddS||j_ dS)zSet the text font :param fontid: the name of a font resource in the Font subdictionary of the current resource dictionary :param fontsize: size is a number representing a scale factor. zUndefined Font id: %rNz Could not set text font because r)rr1rrMrQr rrBrrr rrrN)rWrrN fontsize_fr?r?r@do_Tfs    zPDFPageInterpreter.do_TfrScCr)zSet the text rendering modeNz*Could not set text rendering mode because z is an invalid int value)rrrrrS)rWrSrender_ir?r?r@do_Trs   zPDFPageInterpreter.do_TrrTcCr)z`Set the text rise :param rise: a number expressed in unscaled text space units Nz Could not set text rise because r)r rrrrT)rWrTrise_fr?r?r@do_TsrzPDFPageInterpreter.do_Tstxtyc Cst|}t|}|dur9|dur9|jj\}}}}} } ||||| } ||||| } ||||| | f|j_ntjrGtd|d|dd|j_dS)zlMove to the start of the next line Offset from the start of the current line by (tx , ty). NzInvalid offset (z, z) for Tdr^)r rrIr rrrJ rWrrtx_ty_ar^rrefe_newf_newr?r?r@do_Td s zPDFPageInterpreter.do_Tdc Cst|}t|}|dur9|dur9|jj\}}}}} } ||||| } ||||| } ||||| | f|j_ntjr@td|durH||j_d|j_dS)zMove to the start of the next line. offset from the start of the current line by (tx , ty). As a side effect, this operator sets the leading parameter in the text state. Nz"Invalid offset ({tx}, {ty}) for TDr^)r rrIr rrrRrJrr?r?r@do_TDs zPDFPageInterpreter.do_TDrrrrc CsJ||||||f}t|}|durtd|ddS||j_d|j_dS)z$Set text matrix and text line matrixNz4Could not set text matrix because not all values in rr^)rrrrrIrJ) rWrr^rrrrr;rIr?r?r@do_Tm6s   zPDFPageInterpreter.do_TmcCsL|jj\}}}}}}|||||jj|||jj||f|j_d|j_dS)zMove to start of next text liner^N)rrIrRrJ)rWrr^rrrrr?r?r@do_T_aKs zPDFPageInterpreter.do_T_aseqcCsF|jjdurtjr tddS|j|jtt||j j |j dS)z0Show text, allowing individual glyph positioningNzNo font specified!) rrMr rrBr render_stringr rrror])rWrr?r?r@do_TJXs zPDFPageInterpreter.do_TJscCs||gdS)z Show textN)rrWrr?r?r@do_TjerzPDFPageInterpreter.do_TjcCs|||gdS)zPMove to next line and show text The ' (single quote) operator. N)rrrr?r?r@do__qiszPDFPageInterpreter.do__qawaccCs$||||||gdS)zqSet word and character spacing, move to next line, and show text The " (double quote) operator. N)rrr)rWrrrr?r?r@do__wqs  zPDFPageInterpreter.do__wcCr)zBegin inline image objectNr?rVr?r?r@do_BIzrzPDFPageInterpreter.do_BIcCr)zBegin inline image dataNr?rVr?r?r@do_ID}rzPDFPageInterpreter.do_IDcCs`t|tr*d|vr,d|vr.tt|}|j|dt|j|||j|dSdSdSdS)zEnd inline image objectWHrrrrN) rr#rbidr begin_figurer2 render_image end_figure)rWr\iobjidr?r?r@do_EIs   zPDFPageInterpreter.do_EI xobjid_argc Cs@t|}z t|j|}Wnty tjrtd|YdSwtd|| d}|t urzd|vrz| }t t t|d}t tt| dt}| d}|rXt|} n|j} |j||||j| |gt||jd|j|dS|turd |vrd |vr|j|d t|j|||j|dS dS) zInvoke named XObjectzUndefined xobject id: %rNzProcessing xobj: %rrBBoxr3 ResourcesrWidthHeightr)r1r'rrQr rrBrrr LITERAL_FORMrr r6r%r3r2r$rr]rrrender_contentsr8rr LITERAL_IMAGEr) rWrrxobjr interpreterbboxrIxobjresrr?r?r@do_Dos<        zPDFPageInterpreter.do_DopagecCstd||j\}}}}|jdkrdddd| |f}n'|jdkr*dddd||f}n|jdkr9dddd|| f}n dddd| | f}|j|||j|j|j|d|j |dS) NzProcessing page: %rZrrir) rrmediaboxrotater begin_pagerrcontentsend_page)rWrx0y0r#r$rr?r?r@ process_pages    zPDFPageInterpreter.process_pagercCs6td||||||||t|dS)zTRender the content streams. This method may be called recursively. z1render_contents: resources=%r, streams=%r, ctm=%rN)rrrrexecuter%)rWrrrr?r?r@rs   z"PDFPageInterpreter.render_contentsc Cszt|}Wn tyYdSw z|\}}Wn ty$YdSwt|tr}t|}d|dddddd}t||rqt||}|j j d }|rg| |} t d || t| |krf|| nt d ||ntjr|d |} t| n||q) NTzdo_%s*_a"_w'_qrz exec: %s %rzexec: %szUnknown operator: %r)rr( nextobjectrr,r0replacehasattrgetattr__code__ co_argcountrrrrr rrBr) rWrparserrr\rmethodfuncnargsargsrr?r?r@rsD           zPDFPageInterpreter.execute)rKrr`)dr<r=r>rrprrXrrrrr3r PDFStackTrrrrrrHrcrrrrrr r r rrrrrr r"r0r2r4r7r<r@rCrErFrGrHrIrJrKrLrMrNrRrSrYr[rcrerjrlrxr{r}r~rrrrrrrrrrrrrrrrrrrrrrrrrrrrrr rr2rrrr?r?r?r@rks )                            - -            &  r)Yloggingriortypingrrrrrrr r pdfminerr pdfminer.castingr r rrrpdfminer.cmapdbrrrpdfminer.pdfcolorrrpdfminer.pdfdevicerrpdfminer.pdfexceptionsrrpdfminer.pdffontrrrrrrpdfminer.pdfpager pdfminer.pdftypesr!r"r#r$r%r&r'pdfminer.psexceptionsr(r)pdfminer.psparserr*r+r,r-r.r/r0r1pdfminer.utilsr2r3r4r5r6r7r8 getLoggerr<rr:rBrtrurrrrHfloatColorrcrprrrr?r?r?r@sJ (   $ ($ 7 2Hj