mj{TdZddlZddlZddlZddlmZddlmZddlm Z m Z m Z m Z m Z mZddlmZddlmZddlmZdd lmZdd lmZmZd d lmZd d lmZmZmZd dl m!Z!m"Z"m#Z#d dl$m%Z%m&Z&dZ'dZ(dZ)dZ*dZ+dZ,dZ-dZ.dZ/e*de+de,de-de/de.diZ0ejbdejdZ3e d Z4d!e)d"e*d#e+d$e,d%e-d&e/d'e.d(Z5d)e4d*e4fd+Z6d)e7d*ejfd,Z8d)ejd*e7fd-Z9e6fd.ed/d0d1e7d2e e ge fd*e e:e7f d3Z;Gd4d5eZ<Gd6d0ee&Z=y)7z~ Anything related to Extensible Metadata Platform (XMP) metadata. https://en.wikipedia.org/wiki/Extensible_Metadata_Platform N)Iterator hexdigits)AnyCallableOptionalTypeVarUnioncast)ExpatBuilderNS)Document)Element)Options) ExpatError XMLParserType)XmpInformationProtocol) StreamTypedeprecate_with_replacementdeprecation_no_replacement)LimitReachedError PdfReadErrorXmpDocumentError) ContentStream PdfObjecti@KLiz+http://www.w3.org/1999/02/22-rdf-syntax-ns#z http://purl.org/dc/elements/1.1/zhttp://ns.adobe.com/xap/1.0/zhttp://ns.adobe.com/pdf/1.3/zhttp://ns.adobe.com/xap/1.0/mm/zhttp://ns.adobe.com/pdfx/1.3/zhttp://www.aiim.org/pdfa/ns/id/dcxmppdfxmpMMpdfaidpdfxa{ (?P[0-9]{4}) (- (?P[0-9]{2}) (- (?P[0-9]+) (T (?P[0-9]{2}): (?P[0-9]{2}) (:(?P[0-9]{2}(.[0-9]+)?))? (?PZ|[-+][0-9]{2}:[0-9]{2}) )? )? )? Ku valuereturnc|SNr#s z"_converter_date..s!AQ#a&!As:r)hoursminutes) iso8601match ValueErrorr8groupdecimalDecimal to_integral ROUND_FLOORdatetimesplit timedelta)r#matchesr-r.r0r1r3r4 seconds_decmilliseconds_decseconds millisecondsr5dt tzd_hours tzd_minutess r)_converter_daterStsummE"G0899 w}}V$ %D  g&-# .E gmmE")c *C w}}V$+ ,D x(/C 0F __W]]84; >1 22r+elementselfXmpInformation list_type converterc|jt|}g}t|rN|D]G}|jtdD],}|j|}||}|j |.I|Sy)Nli)getElementsByTagNameNS RDF_NAMESPACElen _get_textappend) r]r^r`ra containersretval containeritemr#s r) _generic_getrms{// yIJF :# %I!88M %t,!%( e$ % %  r+ceZdZdZddeeddffd Zdededeed eed ed eed eeddfd Z dede eddffd Z de ddffd Z xZS) _XmpBuilderac Custom XML parser denying all entity declarations. This is a stripped down and typed version inspired by what *defusedxml* does. Why do we need this? The default limits of *libexpat* used by Python only block exponential entity expansion, but not cases like quadratic entity expansion which can still cause quite some memory usage. Noptionsr$c4t||d|_y)N)rpr)super__init___element_count)r^rp __class__s r)rsz_XmpBuilder.__init__s )r+ entity_nameis_parameter_entityr#base system_id public_id notation_namectd|)NzForbidden entities: )r)r^rvrwr#rxryrzr{s r)!custom_entity_declaration_handlerz-_XmpBuilder.custom_entity_declaration_handlers/ ?@@r+name attributesc|xjdz c_|jtkDrtdtdt|||y)NrzXMP metadata exceeds limit of z elements.)r~r)rtXMP_MAX_ELEMENT_COUNTrrrstart_element_handler)r^r~rrus r)rz!_XmpBuilder.start_element_handlersM q    !6 6#&DEZD[[e$fg g %4J%Gr+parsercht|||j|_|j|_yr&)rrinstallr}EntityDeclHandlerrStartElementHandler)r^rrus r)rz_XmpBuilder.installs+ #'#I#I %)%?%?"r+r&)__name__ __module__ __qualname____doc__rrrsstrboolr}listrrr __classcell__)rus@r)roros  1 T  A A"& AC= A 3- A  A } A$C= A  AH#H49HH @m@@@r+roc eZdZdZdeddfdZedTdZ dUdede de e fddfdZ d e d e d e de efd Zd e d e de efd Zdede fdZefd e d e dee gefdeefdZd e d e deee fdZefd e d e deegefdeeefdZd e d e deeeeffdZedeee fdZej<deee ddfdZedee fdZej<dee ddfdZedeee fdZ e j<deee ddfdZ edeee!jBfdZ"e"j<deee e e!jBfddfdZ"edeee e ffdZ#e#j<deee e fddfd Z#edee fd!Z$e$j<dee ddfd"Z$edee fd#Z%e%j<dee ddfd$Z%edeee fd%Z&e&j<deee ddfd&Z&edeee fd'Z'e'j<deee ddfd(Z'edeee fd)Z(e(j<deee ddfd*Z(edeee e ffd+Z)e)j<deee e fddfd,Z)edee fd-Z*e*j<dee ddfd.Z*edeee fd/Z+e+j<deee ddfd0Z+edeee e ffd1Z,e,j<deee e fddfd2Z,edeee fd3Z-e-j<deee ddfd4Z-edee fd5Z.e.j<dee ddfd6Z.edee fd7Z/e/j<dee ddfd8Z/edee fd9Z0e0j<dee ddfd:Z0edee!jBfd;Z1e1j<dee!jBddfd<Z1edee!jBfd=Z2e2j<dee!jBddfd>Z2edee!jBfd?Z3e3j<dee!jBddfd@Z3edee fdAZ4e4j<dee ddfdBZ4edee fdCZ5e5j<dee ddfdDZ5edee fdEZ6e6j<dee ddfdFZ6edee fdGZ7e7j<dee ddfdHZ7edee fdIZ8e8j<dee ddfdJZ8edeeeffdKZ9dVd e defdLZ:d e d e ddfdMZ;d e d e dee ddfdNZd e d e deee e fddfdQZ?d e de fdRZ@dWdSZAy)Xr_z An object that represents Extensible Metadata Platform (XMP) metadata. Usually accessed by :py:attr:`xmp_metadata()`. Raises: PdfReadError: if XML is invalid streamr$Nc||_ |jj}t|x}tkDrt d|dtdt j |}|jtd}|s td|d|_ i|_ y#ttf$r}td|d}~wwxYw)NzXMP stream size z exceeds limit of .z#XML in XmpInformation was invalid: RDFz?XML in XmpInformation was invalid: Missing rdf:RDF root elementr)rget_datarfXMP_MAX_INPUT_LENGTHrro parseStringAttributeErrorrrrdrerdf_rootcache)r^rdatalengthdoc_roote rdf_rootss r)rszXmpInformation.__init__s  J;;'')Dd)#';;'*:6(BTUiTjjk(lmm!,!:!:4!@H33M5I Q %.aL %'  + J!DQCHI I JsABB=*B88B=crtdd}|jtjd||S)z Create a new XmpInformation object with minimal structure. Returns: A new XmpInformation instance with empty metadata fields. Nutf-8)rset_data _MINIMAL_XMPencode)clsrs r)createzXmpInformation.creates1tT* ++G456{r+encryption_keycptddd| tdd|jj|y)NzXmpInformation.write_to_streamzPdfWriter.xmp_metadataz6.0.0z/the encryption_key parameter of write_to_streamz5.0.0)rrrwrite_to_stream)r^rrs r)rzXmpInformation.write_to_streams> # , $   % &A7  ##F+r+ about_uri namespacer~c#K|jjtdD]N}|jtd|k(s|j ||}|||j||Ed{Py7wN Descriptionabout)rrdregetAttributeNSgetAttributeNodeNS)r^rrr~descattrs r) get_elementzXmpInformation.get_elementsvMM88 V HD""=':iG..y$?#J66y$GGG  H Hs=A7-A7-A5.A7c#K|jjtdD]}|jtd|k(st |j j D]4}|j j|}|s!|j|k(s1|6|jD]}|j|k(s|ywr) rrdrerrangerrrl namespaceURI childNodes)r^rrrirchilds r)get_nodes_in_namespacez%XmpInformation.get_nodes_in_namespacesMM88 V $D""=':iGt556#A??//2D 1 1Y >" #"__$E))Y6# $  $s=B>?B>B>$B>5 B>r]c~d}|jD]+}|j|jk(s||jz }-|SN)rnodeType TEXT_NODEr)r^r]textrs r)rgzXmpInformation._get_textsA'' #E~~0 " # r+rac^|jj|ij|}|r|Sd}|jd||D]:}|j|jk(r |j }n|j |}n|||}|jj|i}|||<|Sr)rgetrrATTRIBUTE_NODE nodeValuerg setdefault)r^rr~racachedr#r]ns_caches r)_get_single_valuez XmpInformation._get_single_value!s  2.2248 M''It< G7#9#99))w/     e$E::((B7 r+c|jj|ij|}|rttt|Sg}|j d||D]F}t ||dx}|j|%|j|}|j|H|jj|i}|||<|S)NrBagr`) rrr rrrrmextendrgrhr) r^rr~rrjr]bagsr#rs r) _getter_bagzXmpInformation._getter_bag7s 2.2248 S 6* *''It< %G$WdeDDQ d#w/ e$  % ::((B7 r+c|jj|ij|}|rttt|Sg}|j d||D]o}t ||d|x}|j|&t ||dx}|j|H||j|} |j| q|jj|i} || |<|S)NrSeq)r`rarr) rrr rrrrmrrgrhr) r^rr~rarrjr]seqsrr#rs r)_get_seq_valueszXmpInformation._get_seq_valuesFs  2.2248 S 6* *''It< %G$WdeyYYf d#&wFF$S d#!$.."9: e$ %::((B7 r+c |jj|ij|}|rttttf|Si}|j d||D]}|j td}t|rH|D]B}|j tdD]'}|j|} | ||jd<)Dl|j||d<|jj|i} || |<|S)NrAltrcxml:langz x-default) rrr dictrrrdrerfrg getAttributer) r^rr~rrjr]altsaltrlr#rs r)_get_langalt_valuesz"XmpInformation._get_langalt_valuescs 2.2248 S#X/ /!#''It< >G11-GD4yFC # : :=$ OF $t 4@Et00<=FF '+nnW&={# >::((B7 r+c.|jtdS)z6Contributors to the resource (other than the authors). contributorr DC_NAMESPACEr^s r)dc_contributorzXmpInformation.dc_contributorus m<  vt <L 4eX%6%67 ''(r+c.|jtdS)z&An unordered array of publisher names. publisherrrs r) dc_publisherzXmpInformation.dc_publishers k::r+c2|jtd|y)Nrrrs r)rzXmpInformation.dc_publishers \;?r+c.|jtdS)zLAn unordered array of text descriptions of relationships to other documents.relationrrs r) dc_relationzXmpInformation.dc_relationrr+c2|jtd|y)Nrrrs r)rzXmpInformation.dc_relationrr+c.|jtdS)z`A language-keyed dictionary of textual descriptions of the rights the user has to this resource.rightsrrs r) dc_rightszXmpInformation.dc_rightss'' h??r+c2|jtd|y)Nrrrs r)rzXmpInformation.dc_rightss   x@r+c.|jtdS)zCUnique identifier of the work from which this resource was derived.sourcerrs r) dc_sourcezXmpInformation.dc_sourcerr+c2|jtd|y)Nrrrs r)r zXmpInformation.dc_sourcerr+c.|jtdS)z\An unordered array of descriptive phrases or keywords that specify the topic of the content.subjectrrs r) dc_subjectzXmpInformation.dc_subjects i88r+c2|jtd|y)Nr#rrs r)r$zXmpInformation.dc_subjectrr+c.|jtdS)z9A language-keyed dictionary of the title of the resource.titlerrs r)dc_titlezXmpInformation.dc_titles'' g>>r+c2|jtd|y)Nr'rrs r)r(zXmpInformation.dc_titles   w?r+c.|jtdS)z@An unordered array of textual descriptions of the document type.typerrs r)dc_typezXmpInformation.dc_types f55r+c2|jtd|y)Nr+rrs r)r,zXmpInformation.dc_types \66:r+c.|jtdS)z:An unformatted text string representing document keywords.Keywordsr PDF_NAMESPACErs r) pdf_keywordszXmpInformation.pdf_keywords%%mZ@@r+c2|jtd|y)Nr/rr1rs r)r2zXmpInformation.pdf_keywords  }j%@r+c.|jtdS)z-The PDF file version, for example 1.0 or 1.3. PDFVersionr0rs r)pdf_pdfversionzXmpInformation.pdf_pdfversions%%m\BBr+c2|jtd|y)Nr8r5rs r)r9zXmpInformation.pdf_pdfversions }lEBr+c.|jtdS)z6The name of the tool that saved the document as a PDF.Producerr0rs r) pdf_producerzXmpInformation.pdf_producerr3r+c2|jtd|y)Nr<r5rs r)r=zXmpInformation.pdf_producerr6r+c8|jtdtS)zYThe date and time the resource was originally created. Returned as a UTC datetime object. CreateDater XMP_NAMESPACErSrs r)xmp_create_datezXmpInformation.xmp_create_date %%m\?SSr+c||r#t|}|jtd|y|jtddy)Nr@r\rrBr^r#date_strs r)rCzXmpInformation.xmp_create_date%3 +E2H  " "=, I  " "=, Er+c8|jtdtS)zTThe date and time the resource was last modified. Returned as a UTC datetime object. ModifyDaterArs r)xmp_modify_datezXmpInformation.xmp_modify_date-rDr+c||r#t|}|jtd|y|jtddy)NrKrFrGs r)rLzXmpInformation.xmp_modify_date2rIr+c8|jtdtS)zjThe date and time that any metadata for this resource was last changed. Returned as a UTC datetime object. MetadataDaterArs r)xmp_metadata_datez XmpInformation.xmp_metadata_date:s%%m^_UUr+c||r#t|}|jtd|y|jtddy)NrOrFrGs r)rPz XmpInformation.xmp_metadata_date?s3 +E2H  " "=.( K  " "=.$ Gr+c.|jtdS)z=The name of the first known tool used to create the resource. CreatorTool)rrBrs r)xmp_creator_toolzXmpInformation.xmp_creator_toolGs%%m]CCr+c2|jtd|y)NrS)rrBrs r)rTzXmpInformation.xmp_creator_toolLs }mUCr+c.|jtdS)zGThe common identifier for all versions and renditions of this resource. DocumentIDrXMPMM_NAMESPACErs r)xmpmm_document_idz XmpInformation.xmpmm_document_idP%%o|DDr+c2|jtd|y)NrWrrYrs r)rZz XmpInformation.xmpmm_document_idU  eDr+c.|jtdS)zZAn identifier for a specific incarnation of a document, updated each time a file is saved. InstanceIDrXrs r)xmpmm_instance_idz XmpInformation.xmpmm_instance_idYr[r+c2|jtd|y)Nr`r]rs r)raz XmpInformation.xmpmm_instance_id^r^r+c.|jtdS)zMThe part of the PDF/A standard that the document conforms to (e.g., 1, 2, 3).partrPDFAID_NAMESPACErs r) pdfaid_partzXmpInformation.pdfaid_partbs%%&6??r+c2|jtd|y)Nrdrrfrs r)rgzXmpInformation.pdfaid_partgs /?r+c.|jtdS)zFThe conformance level within the PDF/A standard (e.g., 'A', 'B', 'U'). conformancerers r)pdfaid_conformancez!XmpInformation.pdfaid_conformanceks%%&6 FFr+c2|jtd|y)Nrkrirs r)rlz!XmpInformation.pdfaid_conformanceps /Fr+ct|dsi|_|jdtD]}|j}d} |j d|}|dk(rn\||dz|dz}t |dk7std |Ds|dz}J|d |tt|d z||dzd z}|dz}t|j|jk(r |j}n|j|}||j|<|jS) z Retrieve custom metadata properties defined in the undocumented pdfx metadata schema. Returns: A dictionary of key/value items for custom metadata properties. _custom_propertiesrruↂr=rc3,K|] }|tvywr&r)r9cs r)r;z3XmpInformation.custom_properties..s2Ra1 >2RsN)rx)hasattrrorPDFX_NAMESPACE localNamefindrfallchrr8rrrrg)r^nodekeystartidxhex_idr#s r)custom_propertiesz XmpInformation.custom_propertiestst12&(D #33BG 5nn((8U3Cby q373F6{a's2R62R/R!$a ds)c#f2*>&??#cAgi.PC!GE==D$7$77 NNE NN40E/4'',) 5*&&&r+c`|jjtdD]}|jtd|k(s|cS|jj}| t d|j td}|jtd||jj||S)zBGet or create an rdf:Description element with the given about URI.rrXMP Document is Nonezrdf:Descriptionz rdf:about) rrdrer ownerDocumentrcreateElementNSsetAttributeNS appendChild)r^rrdocs r)_get_or_create_descriptionz)XmpInformation._get_or_create_descriptionsMM88 V D""=':iG  mm)) ;"#9: :""=2CD M; B !!$' r+cP|jj|}|r ||vr||=yyy)z>), ()8r+c*|j|||j}t|j||}|D]}|j ||j ||x}r|j |||jj}| td|j|} |j|| d|}|jt|} |j| |j||jy)z&Set or remove a single metadata value.Nrr<)rrrrd removeChildrremoveAttributeNoderrr_get_namespace_prefixrcreateTextNoderr_update_stream) r^rr~r#rexisting_elementselem existing_attrrprefix text_nodes r)rz XmpInformation._set_single_values   40..0 !>// :F&&yVHAdV2DED**3u:6I   Y '   T " r+c|j|||j}t|j||}|D]}|j ||r|j j }| td|j|}|j||d|}|jtd} |D]T} |jtd} |jt| } | j| | j| V|j| |j||jy)z+Set or remove bag values (unordered array).Nrr<zrdf:Bagrdf:lirrrrdrrrrrrrerrrr) r^rr~rrrrrrbagr#rcrs r)rzXmpInformation._set_bag_values0  40..0 !>// :F&&yVHAdV2DED%%mY?C $((A..s5z: y)#  $   S !   T " r+c|j|||j}t|j||}|D]}|j ||r|j j }| td|j|}|j||d|}|jtd} |D]T} |jtd} |jt| } | j| | j| V|j| |j||jy)z.Set or remove sequence values (ordered array).Nrr<zrdf:Seqrr) r^rr~rrrrrrseqr#rcrs r)rzXmpInformation._set_seq_valuesrr+c|j|||j}t|j||}|D]}|j ||r|j j }| td|j|}|j||d|}|jtd} |jD]i\} } |jtd} | jd| |jt| } | j| | j| k|j| |j||j!y)z*Set or remove language alternative values.Nrr<zrdf:Altrr)rrrrdrrrrrrreitems setAttributerrrr)r^rr~rrrrrrrlangr#rcrs r)rz"XmpInformation._set_langalt_valuessH  40..0 !>// :F&&yVHAdV2DED%%mY?C%||~ $ e((A D1..s5z: y)#  $   S !   T " r+c.tj|dS)z?Get the appropriate namespace prefix for a given namespace URI.unknown)_NAMESPACE_PREFIX_MAPr)r^rs r)rz$XmpInformation._get_namespace_prefixs$((I>>r+c|jj}| td|jd}|jj |y)z/Update the stream with the current XML content.Nrr)encoding)rrrtoxmlrr)r^rxml_datas r)rzXmpInformation._update_streamsDmm)) ;"#9: :99g9. X&r+)r$r_r&)r)r$N)Brrrrrrs classmethodrrr rbytesrrrrr XmlElementrgr*rrrrrrrrpropertyrsetterrrrHrrrr rrrrr r$r(r,r2r9r=rCrLrPrTrZrargrlrrrrrrrrrr'r+r)r_r_s} (}(("  MQ ,  ,27c58H2I ,  ,HSHSHHQT H $ $ $QT $+4 SE3J'  # , S  c8K &+4 SE3J'  $s)  :ScSVh@X$=c 3==BXd3i%8BTBB@Xc]@@@#@4@@=HT#Y/==>$s)!4>>>K$x'8'8"9:KK ^^ EhtE#x7H7H2H,I'JK EPT E EEc3h 8EEFXd38n%=F$FF>8C=>>>x}>>>Bx}BBB8C=BTBB:Xd3i0::?(49"5?$??;htCy1;;@8DI#6@4@@:Xd3i0::?(49"5?$??@8DcN3@@Ac3h 8ATAA>8C=>>>x}>>>9HT#Y/99>$s)!4>>>?(4S>2??__@xS#X7@D@@6$s),66 ^^;htCy1;d;;AhsmAAA(3-ADAAC CCCHSMCdCCAhsmAAA(3-ADAAT(*;*;!<TTFXh.?.?%@FTFFT(*;*;!<TTFXh.?.?%@FTFFV8H,=,=#>VVHx0A0A'BHtHHD(3-DDDhsmDDDE8C=EEEx}EEEE8C=EEEx}EEE@Xc]@@@#@4@@GHSMGGG G$GG '4S> ' 'D C  Cst 3c(3-TX0C$s)ATY]8C$s)ATY]8SXdSVX[S[nE]bf:?s?s?'r+)>rrHrDrecollections.abcrstringrtypingrrrr r r xml.dom.expatbuilderr xml.dom.minidomr rrxml.dom.xmlbuilderrxml.parsers.expatrr _protocolsr_utilsrrrerrorsrrrgenericrrrrrerrBr1rYrvrfrcompileVERBOSEr@r"rr*rrSr\rrmror_r'r+r)rs  $0$1&7.VVEE- = 1 . . 3(15$55WhF  "** JJ! ( CL&(#n%%'%')*++,-'()  Q138#4#46 3 1 1 3c 3hq  #3 @C PXZ]Y^`cYcPd  d3i $@.$@NQ '+YQ 'r+