j- SrSSKrSSKrSSKrSSKJr SSKJr SSKJ r J r SSK J r SSK r SSKJrJrJr SSKJr SS KJr SS KJr S rS r"S S\5rS\S\4SjrS\S\S-4SjrS S\S\ \\4S\4Sjjr!S!S\\-\-S\ \\4S-S\"S-S\4Sjjr#S!S\\-\-S\ \\4S-S\S-S\4Sjjr$S!S\\-\-S\ \\4S-S\S-S\4Sjjr%\"S5S"S\\-\-S\ \\4S-S\4Sjj5r&g)#zFile-related utilities.N)BytesIO)Path)OptionalUnion)urlparse) AnyHttpUrl TypeAdapterValidationError) deprecated) relative_path)DocumentStreamic<^\rSrSrSrS\S\S\4U4SjjrSrU=r $)FileSizeLimitExceededErrorzERaised when a remote file exceeds the configured download size limit.filenamesizelimitcV>XlX lX0l[TU]SUSUS35 g)Nz.Remote file exceeds the maximum allowed size (z > z bytes).)rrrsuper__init__)selfrrr __class__s T/home/mande/repo/quber/.venv/lib/python3.13/site-packages/docling_core/utils/file.pyr#FileSizeLimitExceededError.__init__s2    I$sSXRYYabc)rrr) __name__ __module__ __qualname____firstlineno____doc__strintr__static_attributes__ __classcell__)rs@rrrs*OddCdddrrurlreturncT[U5nURnU(dg[R"U5nUR=(ap UR=(dX UR=(dE UR=(d2 UR=(d UR=(d UR(+$![aR SSKnUR U5n[R"U5nN!URUR4a gf=ff=f![ a gf=f)z=Return whether a URL resolves to a globally routable address.FrN)rhostname ipaddress ip_address ValueErrorsocket gethostbynamegaierrorherror is_global is_private is_loopback is_link_local is_reserved is_multicastis_unspecified Exception)r&parsedr)ipr-ip_strs r _is_safe_urlr<$s#?? %%h/B|| MM !~~ ! !~~ !  !   !     --h7))&1OOV]]3    " sRDB;BD;D 'C31D3DDDDDD D'&D'rc|URSS5n[U5RnU(a US;dSU;agU$)zGReturn a basename-safe filename, or None if no usable basename remains.\/).z..N)replacername)r normalizedbasenames r_sanitize_filenamerEDs<!!$,JJ$$H x;.#/ Orhttp_urlresponse_headersc*SnURS5=n(aUR5RS5HpnURS5=n(dMUSR5S:XdM6SRUSS5R5RS5=(d Sn O Uc1[ UR =(d S 5R =(d Un[U5=n(aU$[U5=n(aU$[S 5e) aFResolves the filename from a remote url and its response headers. Args: source AnyHttpUrl: The source http url. response_headers Dict: Headers received while fetching the remote file. fallback_filename str: Filename to use in case none can be determined. Returns: str: The actual filename of the remote url. NzContent-Disposition;=rrz'"z Could not derive a safe filename) getstripsplitjoinrpathrBrEr,)rFrGfallback_filename raw_fname cont_dispparrOfnames rresolve_remote_filenamerWOsI$(()>??y???$**3/C3''U1X^^-=-KHHU12Y/557==eDL 0 ,"-22G6G "9--u- "#455u5 7 88rsourceheaders max_file_sizec [[5RU5n[U5n[ U5(d[ SU35eU=(d 0nUR 5VVs0sHupgUR5U_M nnnSU;a'SSKn SU RRS53n XS'[R"SU5n U (aU RS5n U RS 5n U S :XaS U 3nO5U S :XaS U S3nO(U S:XaSU S3nOU S:XaSU S3nO[ SU 35e[[5RU5nSn[R "5n["UlUR&SR)U5 UR+USUSS9nUR-5 [/UR05n[3UUS9nUb>UR0R+S5nUb [5U5nUbUU:a [7UUUS9e[95nSnUR;[<S9H?nU(dM U[?U5- nUbUU:a [7UUUS9eURAU5 MA URCS5 [EUUS9nSSS5 SSS5 W$s snnf![a Sn GN f=f![ a SnNf=f!,(df  NC=f!,(df  W$=f![Fa [IU[5(a>SU;a8URKSS5SR5nUS ;a[ S!US"35e[[L5RU5n[9URO55n[EURPUS9nU$![Fa [ S#[SU535ef=ff=f)$aResolves the source (URL, path) of a file to a binary stream. Args: source: The file input source. Can be a path or URL. headers: Optional set of headers to use for fetching the remote URL. max_file_size: Optional maximum size, in bytes, for a remote download. When set, the download is rejected upfront if the declared ``Content-Length`` exceeds it, and aborted while streaming as soon as the received bytes exceed it. Raises: ValueError: If source is of unexpected type. FileSizeLimitExceededError: If a remote download exceeds ``max_file_size``. Returns: DocumentStream: The resolved file loaded as a stream. zURL is not allowed: z user-agentrNz docling-core/z docling-corezCgoogle\.com\/(file|document|spreadsheets|presentation)\/d\/([\w-]+)rKfilez/https://drive.google.com/uc?export=download&id=documentz#https://docs.google.com/document/d/z/export?format=docx spreadsheetsz'https://docs.google.com/spreadsheets/d/z/export?format=xlsx presentationz'https://docs.google.com/presentation/d/z/export?format=pptxzUnexpected Google doc type: c*UR(dUR(aqURRS5nU(aNUR S5(dSSKJn U"URU5n[U5(d[SU35eggg)z2Validate each redirect target before following it.location)zhttp://zhttps://r)urljoinz Redirect target is not allowed: N) is_redirectis_permanent_redirectrYrM startswith urllib.parsercr&r<r,)responseargskwargs redirect_urlrcs r_check_redirect_safety8resolve_source_to_stream.._check_redirect_safetys##x'E'E'//33J? '223JKK8'.x||\'J ' 55(+KL>)Z[[6  (FrrhT)streamrYallow_redirects)rFrGzContent-Length)rrr) chunk_size)rBrnz://)httphttpszUnsupported URL scheme: 'z+'. Only http:// and https:// are supported.z$Unexpected source type encountered: )*r rvalidate_pythonr"r<r,itemslowerimportlib.metadatametadataversionr8researchgrouprequestsSession_MAX_REDIRECTS max_redirectshooksappendrMraise_for_statusdictrYrWr#rr iter_content_DOWNLOAD_CHUNK_SIZElenwriteseekr r isinstancerOr read_bytesrBtype)rXrYrZrFurl_str_headerskv req_headers importlib agent_name google_doc_iddoc_typedoc_idrlsessionresrGrVcontent_lengthcontent_length_valuern downloadedchunk doc_streamscheme local_paths rresolve_source_to_streamrqs,nT*:6FFvNh-G$$3G9=> >=b080@A0@qwwy!|0@ A { * ,),Y-?-?-G-G-W,XY )3 % R   $**1-H"((+F6!KF8TZ'?xGZ[^+CF8K^_^+CF8K^_ #?z!JKK":.>>wGH \    7$2G ! MM* % , ,-C D# $  $$&#' #4 /Tde ,%([[__5E%FN%1836~3F00;@TWd@d"<).%9&3# !  --9M-NE  #e*,J$0Z-5O8%*!+"/ LL'O A+vF Q  p OB  ,+  ,z *83708!  p   T fc " "u\\%+A.446F.. #RS S T TsA%K5'J' K5"J-0B;K5+rs "!==(60 "ddcd@t9938n9 9H&* $E :  #E #s(^d "E:E ET&* :  # #s(^d "D[ 0&* :  # #s(^d "D[ < UV&* :  # #s(^d " Wr