h5j'hUdZddlmZddlZddlmZddlmZmZm Z ddl m Z ddl m Z mZmZerddl mZd gZid d d d d ddddddddddddddddddddd d!d"d#d$d%d%d&d'id(d)d*d+d,d-d.d/d0d1d2d2d3d4d5d6d7d8d9d:d;d;dd?d@dAdBdBdCdDdEdFidGdHdIdJdKdLdMdNdOdPdQdRdSdTdUdVdWdXdYdZd[d\d]d]d^d_d`dadbdcdddedfdgidhdidjdkdldldmdndodpdqdrdsdtdudvdwdxdydzd{d|d}d~ddddddddddiddddddddddddddddddddddddddddddddddZded<ehdZehdZeej/ezezZdZddZddZddZGdd Z ddZ ddZy)aMA Pythonic builder for qpdf jobs. qpdf exposes a large amount of functionality through its "job" interface: the same operations the ``qpdf`` command line tool performs (encrypt, decrypt, merge/split pages, linearize, compress, manage attachments, overlay/underlay, and so on). pikepdf binds this as :class:`pikepdf.Job`, which accepts a job specified as a JSON string, a Python ``dict``, or command line arguments. Writing that JSON by hand is awkward: the keys are camelCase, the encryption block has ``40bit``/``128bit``/``256bit`` variants with their own restriction vocabulary, and permission flags are expressed as restrictions rather than grants. :class:`JobBuilder` provides a fluent, snake_case, Pythonic API that assembles a valid job specification and hands it to :class:`pikepdf.Job`. qpdf's JSON job format is a complete representation of any job configuration -- ``initializeFromJson``, ``initializeFromArgv`` and qpdf's own fluent C++ ``config()`` builder all funnel through the same handlers -- so nothing is lost by building the specification in Python. Example: >>> from pikepdf import JobBuilder, Permissions >>> ( ... JobBuilder() ... .input('in.pdf') ... .output('out.pdf') ... .encrypt(owner_password='secret', allow=Permissions(extract=False)) ... .run() ... ) # doctest: +SKIP ) annotationsN)deepcopy) TYPE_CHECKINGAnyLiteral)Job)DEFAULT_PERMISSIONS Encryption Permissions)Pdf JobBuilder input_file inputFilepassword password_file passwordFileempty json_input jsonInput output_file outputFile replace_input replaceInputqdfpreserve_unreferencedpreserveUnreferencednewline_before_endstreamnewlineBeforeEndstreamnormalize_contentnormalizeContent stream_data streamDatacompress_streamscompressStreamsrecompress_flaterecompressFlate decode_level decodeLeveldecryptdeterministic_iddeterministicId static_aes_iv staticAesIv static_idstaticIdno_original_object_idsnoOriginalObjectIdscopy_encryptioncopyEncryptionencryption_file_passwordencryptionFilePassword linearizelinearize_pass1linearizePass1object_streams objectStreams min_version minVersion force_version forceVersionprogress split_pages splitPages json_output jsonOutputremove_restrictionsremoveRestrictionscheckcheck_linearizationcheckLinearizationfiltered_stream_datafilteredStreamDataraw_stream_data rawStreamDatashow_encryptionshowEncryptionshow_encryption_keyshowEncryptionKeyshow_linearizationshowLinearization show_npages showNpages show_object showObject show_pages showPages show_xrefshowXref with_images withImageslist_attachmentslistAttachmentsshow_attachmentshowAttachmentjsonjson_stream_datajsonStreamDatajson_stream_prefixjsonStreamPrefixupdate_from_jsonupdateFromJsonallow_weak_cryptoallowWeakCryptokeep_files_open keepFilesOpenkeep_files_open_thresholdkeepFilesOpenThresholdno_warnnoWarnverbosetest_json_schematestJsonSchemaignore_xref_streamsignoreXrefStreamspassword_is_hex_keypasswordIsHexKey password_mode passwordModesuppress_password_recoverysuppressPasswordRecoverysuppress_recoverysuppressRecoverycoalesce_contentscoalesceContentscompression_levelcompressionLevel jpeg_quality jpegQualityexternalize_inline_imagesexternalizeInlineImages ii_min_bytes iiMinBytesremove_unreferenced_resourcesremoveUnreferencedResourcescollateflatten_annotationsflattenAnnotationsflatten_rotationflattenRotationgenerate_appearancesgenerateAppearanceskeep_inline_imageskeepInlineImages oi_min_area oiMinArea oi_min_height oiMinHeight oi_min_width oiMinWidthoptimize_imagesoptimizeImagesremove_acroformremoveAcroform remove_info removeInforemove_metadataremoveMetadataremove_page_labelsremovePageLabelsremove_structureremoveStructurereport_memory_usagereportMemoryUsage warning_exit0 warningExit0preserve_unreferenced_resourcespreserveUnreferencedResourcesrequires_passwordrequiresPassword is_encrypted isEncrypteddict[str, str]_KEY_MAP>form40bitprint128bit256bitmodifyuseAesencryptextractforceR5forceV4annotateassemble modifyOther userPassword accessibility allowInsecure ownerPasswordcleartextMetadata>tokeyfilefrompagesrangeglobalprefixrepeatrotatemoddateoverlayreplacefilenamemimetypeunderlay description creationdate addAttachment setPageLabelsnoDefaultLimitsparserMaxErrorsmaxStreamFiltersparserMaxNestingremoveAttachmentcopyAttachmentsFromparserMaxContainerSizeparserMaxContainerSizeDamagedct|tjrtj|}t|trtj |St |S)z9Coerce a path-like to the str form qpdf job JSON expects.) isinstanceosPathLikefspathbytesfsdecodestr)ps B/var/www/html/qr/venv/lib/python3.12/site-packages/pikepdf/jobs.py_pathrs>!R[[! IIaL!U{{1~ q6Mc|rdSdS)Nyn)alloweds r_ynrs3"s"rc(|dk\ry|dk(ry|dk(ryy)zEMap an :class:`Encryption` security-handler revision to a key length.(r)Rs r _bits_for_Rrs#AvAvAv rc eZdZUdZded<dBdZdddCdZdDdZdEd ZdDd Z dFddd dd dd d d d dGdZ dDdZ dDdZ dDdZ dFdd dHdZdFdIdZdFdJdZdFdKdZdDdZdDdZddd dddd dLdZdddd dd dMdZdddNdZ dO dPdZdDd ZdDd!ZdDd"ZdDd#ZdDd$ZdDd%ZdDd&ZdDd'ZdDd(Z dQd)Z!dRd*Z"dRd+Z#dDd,Z$dDd-Z%d d.dSd/Z&ddddddd d0 dTd1Z'ddd2 dUd3Z(dVd4Z)ddddd5 dWd6Z*ddddd5 dWd7Z+ddddddd8 dXd9Z,dYd:Z-dZd;Z.d[d<Z/d\d=Z0d d>d]d?Z1d^d@Z2d[dAZ3y)_r a^Fluently assemble a qpdf job and run it. Each method records part of the job specification and returns ``self`` so calls can be chained. List-valued sections (pages, attachments, overlay/underlay) use repeatable ``add_*`` methods. Terminal methods :meth:`build`, :meth:`run` and :meth:`create_pdf` hand the assembled specification to :class:`pikepdf.Job`. The builder performs minimal local validation; qpdf is the source of truth and will raise :class:`pikepdf.JobUsageError` (or ``RuntimeError`` for malformed JSON) for invalid configurations when the job is built. dict[str, Any]_specci|_y)z"Create an empty job specification.Nrselfs r__init__zJobBuilder.__init__s  rN)rcd|jvr tdt||jd<|||jd<|S)zSet the input file, optionally with a password. Args: file: Path to the input PDF. password: Password for an encrypted input file. rz&Cannot set an input file after empty()rr)r ValueErrorr)rrrs rinputzJobBuilder.inputsG djj EF F"'+ ;  %-DJJz " rc^d|jvr tdt|jd<|S)z3Use an empty PDF as input instead of an input file.rz,Cannot use empty() when an input file is setr)rr_ENABLErs rrzJobBuilder.empty s- $** $KL L% 7 rc6t||jd<|S)zSet the output file. Args: file: Path to write. For :meth:`split_pages`, include a ``%d`` placeholder for the page group number. r)rr)rrs routputzJobBuilder.outputs$); <  rc,t|jd<|S)z4Overwrite the input file with the output (in place).rrrrs rrzJobBuilder.replace_inputs%, >" rrTF) owner_password user_passwordbitsallowmetadataaesforce_v4allow_insecureforce_r5c ~|qtd|||fDr td|j}|j}|j}|j }|j }t|j}|dvr td|t}|d} d|xsdd|xsd| t|||||| | i|jd <|S) aEncrypt the output. Either pass a :class:`pikepdf.Encryption` object positionally, or use the keyword arguments. The two forms are mutually exclusive. Args: encryption: A :class:`pikepdf.Encryption` describing passwords, permissions and algorithm. If given, the keyword arguments must not be used. owner_password: Owner password (full access). user_password: User password (restricted access). bits: Key length: 40, 128 or 256 (default). 40 and 128-bit RC4 are weak and require :meth:`allow_weak_crypto`. allow: A :class:`pikepdf.Permissions` describing what the user password is permitted to do. Permissions default to :data:`pikepdf.models.encryption.DEFAULT_PERMISSIONS`. metadata: If False, document metadata is left unencrypted (128/256-bit only). aes: For 128-bit, request AES rather than RC4. force_v4: For 128-bit, force ``V=4`` in the encryption dictionary. allow_insecure: For 256-bit, allow an insecure empty owner password. force_r5: For 256-bit, use the deprecated ``R=5`` algorithm. c3$K|]}|du ywNr).0vs r z%JobBuilder.encrypt..IsQQ1D=Qsz?Pass either an Encryption object or keyword arguments, not both)rrrzbits must be 40, 128 or 256bitrrr)rrrrrr) anyrowneruserrrrrrr _encrypt_flagsr) r encryptionr r rrrrrrrvariants rrzJobBuilder.encrypt#sJ  !Q>=%*PQQ U(--N&OOM$$E!**H..Cz||,D ~ %:; ; ='EF#, M/R ^1r ^!!-! !  9 rc,t|jd<|S)z&Remove encryption from the input file.r)r rs rr)zJobBuilder.decryptis ' 9 rc,t|jd<|S)zFRemove security restrictions, recovering the encryption key if needed.rEr rs rrDzJobBuilder.remove_restrictionsns+2 '( rc,t|jd<|S)z2Permit writing files with weak (RC4) cryptography.rir rs rrhzJobBuilder.allow_weak_cryptos(/ $% rcdt|i}|||d<|||d<|jjdgj||S)abAppend pages from a file to the page-selection (merge/split) operation. Args: file: Source PDF. Use ``'.'`` to refer to the primary input file. page_range: qpdf page range, e.g. ``'1-5'`` or ``'z-1'`` (reversed). Omit to use all pages. password: Password for an encrypted source file. rrrrrr setdefaultappend)rr page_rangerentrys r add_pageszJobBuilder.add_pageszsU"(t 5  !'E'N   (E*  gr*11%8 rc>|dn t||jd<|S)zWrite pages to separate files. Args: group: Number of pages per output file. The output filename should contain a ``%d`` placeholder. rrArr)rgroups rr@zJobBuilder.split_pagess!*/2CJ <  rc>|dn t||jd<|S)zCollate rather than concatenate the page selection. Args: n: Number of pages to take from each file per round. rrr.)rrs rrzJobBuilder.collates! '(iSV 9 rc| t|n|d|}|jjdgj||S)aRotate pages. Args: angle: Rotation in degrees: ``90``, ``180``, ``270``; prefix with ``+`` or ``-`` to rotate relative to the current rotation. page_range: Page range to rotate. Omit to rotate all pages. :r)rrr(r))rangler*specs rrzJobBuilder.rotatesC(/s5zwa |5L h+2248 rc,t|jd<|S)z$Linearize (web-optimize) the output.r6r rs rr6zJobBuilder.linearizes") ; rc,t|jd<|S)zs rrz$JobBuilder.externalize_inline_imagess0 18 ,-  '*9~DJJ| $ rc$||jd<|S)aPush annotations into page content streams. Args: mode: Which annotations to flatten: ``'all'`` (default), ``'print'`` (only those marked for printing) or ``'screen'`` (exclude those marked hidden on screen). rr)rmodes rrzJobBuilder.flatten_annotationss,0 '( rc,t|jd<|S)z;Bake each page's ``/Rotate`` value into its content stream.rr rs rrzJobBuilder.flatten_rotationr%rc,t|jd<|S)z;Generate appearance streams for form fields that lack them.rr rs rrzJobBuilder.generate_appearancess,3 () rc,t|jd<|S)z3Combine a page's multiple content streams into one.r~r rs rr}zJobBuilder.coalesce_contents)0 %& rc,t|jd<|S)zCNormalize newlines in content streams (for readability/inspection).r r rs rrzJobBuilder.normalize_content#rErc,t|jd<|S)z*Remove the document's XMP metadata stream.rr rs rrzJobBuilder.remove_metadata*'. #$ rc,t|jd<|S)z+Remove the document information dictionary.rr rs rrzJobBuilder.remove_info/s#* <  rc,t|jd<|S)z2Remove the interactive form (AcroForm) dictionary.rr rs rrzJobBuilder.remove_acroform4rHrc,t|jd<|S)z/Remove the document's structure (tagging) tree.rr rs rrzJobBuilder.remove_structure9r%rc,t|jd<|S)z-Remove explicit page labels (page numbering).rr rs rrzJobBuilder.remove_page_labels>rErc6t||jd<|S)aSet page labels (explicit page numbering) for the whole document. Args: labels: A list of qpdf label specs of the form ``first-page:[type][/start[/prefix]]``, e.g. ``['1:r', '5:D/1']`` to number the first four pages with lower-case Roman numerals and restart with Arabic numerals from page 5. The first spec must start at page 1. r)listr)rlabelss rset_page_labelszJobBuilder.set_page_labelsEs'+6l ?# rc$||jd<|S)zSet the minimum PDF version of the output. Args: version: A PDF version such as ``'1.7'``, optionally with an extension level as ``'1.7:8'``. r<rrversions rr;zJobBuilder.min_versionTs$+ <  rc$||jd<|S)zForce the output PDF version, even if features require a higher one. Args: version: A PDF version such as ``'1.7'``, optionally with an extension level as ``'1.7:8'``. r>rrRs rr=zJobBuilder.force_version^s&- >" rc,t|jd<|S)zDGenerate the document ID deterministically from the output contents.r+r rs rr*zJobBuilder.deterministic_idjr%rc,t|jd<|S)zAUse a fixed document ID (for testing; not for production output).r/r rs rr.zJobBuilder.static_idos!( : r) linearizationcVt|jd<|rt|jd<|S)zCheck the PDF for problems, reporting via the job's output. Args: linearization: Also check the linearization (hint) tables. rFrHr )rrWs rrFzJobBuilder.checkts) & 7 /6DJJ+ , r)rrrrrrrcdt|i} ||| d<||| d<||| d<||| d<||| d<||| d<|r t| d<|jjd gj | |S) aAttach (embed) a file in the output. Args: file: Path to the file to attach. key: Attachment key; defaults to the filename. filename: Displayed filename of the attachment. mimetype: MIME type, e.g. ``'application/pdf'``. description: Human-readable description. creationdate: Creation date (PDF date string). moddate: Modification date (PDF date string). replace: Replace an existing attachment with the same key. rrrrrrrrr)rrrr(r)) rrrrrrrrrr+s radd_attachmentzJobBuilder.add_attachments0"(t 5 ?E%L   (E*    (E*   "#.E-  #$0E. !  &E)  &E)  or299%@ r)rrcdt|i}|||d<|||d<|jjdgj||S)aCopy all attachments from another PDF. Args: file: Source PDF to copy attachments from. prefix: Prefix to disambiguate keys that collide with existing ones. password: Password for an encrypted source file. rrrrr')rrrrr+s rcopy_attachments_fromz JobBuilder.copy_attachments_fromsV"(t 5  $E(O   (E*  3R8??F rc\|jjdgj||S)zbRemove an embedded file by key. Args: key: Attachment key to remove. r)rr(r))rrs rremove_attachmentzJobBuilder.remove_attachments) 0"5< ``noWarn``). A boolean ``True`` enables a flag (emitted as an empty string); other values are stringified. Args: **kwargs: Option names and values. Raises: ValueError: If a keyword does not correspond to a known option. zUnknown job option: NTF)itemsrKeyErrorrrr)rkwargsnamevaluejson_keys rsetzJobBuilder.sets"<<> -KD% L#D>}'. 8$%', 8$ -  L #7x!@AtK Ls AA*c,t|jS)z@Return a deep copy of the assembled job specification as a dict.)rrrs rto_jsonzJobBuilder.to_json5s ##rc@ddl}|j|jS)z8Return the assembled job specification as a JSON string.rN)radumpsr)rras r to_json_strzJobBuilder.to_json_str9szz$**%%rc,t|jS)a<Construct a :class:`pikepdf.Job` from the specification. The job is validated by qpdf during construction but not executed. Raises: pikepdf.JobUsageError: If the configuration is semantically invalid. RuntimeError: If the JSON is malformed or contains unknown keys. )rrrs rbuildzJobBuilder.build?s4::r)validatecj|j}|r|j|j|S)aaBuild and run the job. Args: validate: If True (default), call ``check_configuration()`` before running to fail fast on invalid configurations. Returns: The :class:`pikepdf.Job` after running, so callers can inspect ``exit_code``, ``has_warnings`` and ``encryption_status``. )r~check_configurationrun)rrjobs rrzJobBuilder.runJs,jjl   # # %   rc>|jjS)a3Build the job and run only its first stage, returning a :class:`pikepdf.Pdf`. This is the staged workflow: the returned PDF can be modified before calling :meth:`pikepdf.Job.write_pdf` on the same job. Use :meth:`build` to retain a reference to the job for the write stage. )r~ create_pdfrs rrzJobBuilder.create_pdf[szz|&&((rc"d|jdS)Nzrrs r__repr__zJobBuilder.__repr__ds%djj^155r)returnNone)rrr str | Nonerr )rr )rrrr r)r zEncryption | Noner rr rrzLiteral[40, 128, 256]rzPermissions | Nonerboolr bool | Nonerrrrrrrr )rrr*rrrrr )r/ int | Nonerr )rrrr )r3z int | strr*rrr )r#rr9z1Literal['generate', 'preserve', 'disable'] | Noner%rrrr'z;Literal['none', 'generalized', 'specialized', 'all'] | Noner!z4Literal['compress', 'preserve', 'uncompress'] | Nonerr ) r9rr:rr;rrrrrrr )r>rrr )all)rAz!Literal['all', 'print', 'screen']rr )rOz list[str]rr )rSrrr )rWrrr )rrrrrrrrrrrrrrrrrr )rrrrrrrr )rrrr ) rrrrr_rrrrrrr )rgrrhrrirrjrrkrrlrrr )rsrrr )rr)rr)rr)rrrr)rr )4__name__ __module__ __qualname____doc____annotations__rrrr rrr)rDrhr,r@rrr6rr8rrrrrr}rrrrrrrPr;r=r*r.rFrZr\r^rdrfrorwryr|r~rrrrrrr r s   :> )-D&*$(&)$($D%D# D " D $ D"DDDDDD DL  37RV%/EO &  )-LP!&(,LP!&!J !  ! & !!J! !L $ $!%#(#'    ! ! >DH 9> 5             .3 "##"&#'"(( (  (  ( (!((( (V26d$.AK $ !#       6 !#       .*.048<(,)-)-"'"$. " ,6 " & "'"'" "L8$& '+")6rcXdt|i}|||d<|||d<|||d<|||d<|S)Nrrrrr)r)rrr_rrr+s rrbrbhsS$U4[1E ~d  f   h$j Lrc i}|jsd|d<n|jsd|d<nd|d<|dk(rt|j|d<t|j|d<|j rd|d <|S|j rd |d <|S|jrd|d <|S|jrd |d <|Sd|d <|St|j|d <t|j|d<t|j|d<t|j|d <t|j |d <t|j |d<|s t|d<|dk(r|rd|d<|r t|d<|S|dk(r|r t|d<|r t|d<|S)aTranslate a :class:`Permissions` grant into qpdf restriction flags. pikepdf's :class:`Permissions` is allow-oriented; qpdf's flags are restrictions. Every flag is emitted explicitly so the result is fully determined by ``allow`` rather than qpdf's defaults. nonerlowfullrrrrrrassemblyrrrrrrrrrrr) print_lowres print_highresrrmodify_annotation modify_other modify_formmodify_assemblyrr)rrrrrrrflagss rrr{s E   g  gg rzu}}-i 7 78j   #E(O   $E(O  $ $(E(O   " "(E(O %E(O !!4!45E/5==)E)E334E*E112E*))*E&Mu112E- %,!" s{ !E(O &E)  L  %,E/ " &E)  Lr)rrrr)rrrzLiteral['y', 'n'])rintrr) rrrrr_rrrrrrr)rrrr rrrrrrrrrrrr) r __future__rrcopyrtypingrrr pikepdf._corerpikepdf.models.encryptionr r r r __all__rr frozenset _ENCRYPT_KEYS _BLOCK_KEYSvalues ALL_JSON_KEYSrrrrr rbrrrrrs<# ..RR! .V+V V^V W V + V < V^V 5V3V 8V+V<V)V)VMV y!V")#V$]%V&'V(3)V*'+V, 8-V./V0'1V2o3V4<5V6^7V8 9V:<;V<<=V>/?V@ WAVB/CVD0EVFGVH'IVJ.KVL-MVN