Mi0Cj%SrSSKJr SSKrSSKrSSKrSSKrSSKrSSK r SSK J r J r SSK Jr SSKJrJrJrJrJr SSKJr SSKJrJr SS KJr SS KJr SS KJr SS K J!r! SS K"J#r# SSK$J%r% Sr&S\'S'SSSSSSS.SSjjjr("SS5r)"SS5r*"SS5r+g)z{Provides `partition_email()` function. Suitable for use with `.eml` files, which can be exported from many email clients. ) annotationsN) EmailMessageMIMEPart)cached_property)IOAnyFinalIteratorcast)parser)ElementElementMetadata)FileType)logger)EXPECTED_ATTACHMENT_ERRORS)get_last_modified_date)partition_html)partition_text) text/html text/plainzFinal[tuple[str, ...]]VALID_CONTENT_SOURCESrT)filecontent_sourcemetadata_filenamemetadata_last_modifiedprocess_attachmentsc l[RUUUUUUUS9n[[R US95$)aPartitions an .eml file into document elements. Args: filename: str path of the target file. file: A file-like object open for reading bytes (not str) e.g. --> open(filename, "rb"). content_source: The preferred message body. Many emails contain both a plain-text and an HTML version of the message body. By default, the HTML version will be used when available. Specifying "text/plain" will cause the plain-text version to be preferred. When the preferred version is not available, the other version will be used. metadata_filename: The file-path to use for metadata purposes. Useful when the target file is specified as a file-like object or when `filename` is a temporary file and the original file-path is known or a more meaningful file-path is desired. metadata_last_modified: The last-modified timestamp to be applied in metadata. Useful when a file-like object (which can have no last-modified date) target is used. The last-modified metadata is otherwise drawn from the filesystem when a path is provided. process_attachments: When True, also partition any attachments in the message after partitioning the message body. All document elements appear in the single returned element list. The filename of the attachment, when available, is used as the `filename` metadata value for elements arising from the attachment. Note that all global keyword arguments such as `unique_element_ids`, `language` and `chunking_strategy` can be used and will be passed along to the decorators that implement those functions. Further, any keyword arguments applicable to HTML will be passed along to the HTML partitioner when processing an HTML message body.  file_pathrrmetadata_file_pathrrkwargsctx)EmailPartitioningContextloadlist_EmailPartitioner iter_elements)filenamerrrrrr!r#s v/var/www/eduai.edurigo.com/storigo/production/storigo_env/lib/python3.13/site-packages/unstructured/partition/email.pypartition_emailr+sKF # ' ' %,5/ ( C !//C/8 99c\rSrSrSrSSSSSS04SSjjr\SSj5r\SSj5r \SS j5r \SS j5r \SS j5r \SS j5r \S S j5r\S Sj5r\S Sj5r\S Sj5r\S!Sj5r\S"Sj5r\S#Sj5r\S Sj5r\SSj5r\S Sj5r\S Sj5rS$SjrSrg)%r$OzVEncapsulates partitioning option validation, computation, and application of defaults.NrFcXXlX lX0lX@lXPlX`lXplgN) _file_path_file_content_source_metadata_file_path_metadata_last_modified_process_attachments_kwargs)selfrrrr rrr!s r*__init__!EmailPartitioningContext.__init__Rs+$ -#5 '=$$7! r,c 6U"UUUUUUUS9R5$)z#Construct and validate an instance.r) _validate)clsrrrr rrr!s r*r%EmailPartitioningContext.loadds/)1#9 3  )+ r,cURRS5nU(dg[RR U5nUVs/sH"n[RR U5PM$ sn$s snf)z6The "blind carbon-copy" Bcc: addresses of the message.BccNmsgget_allemailutils getaddresses formataddr)r8bccsaddrsaddrs r* bcc_addresses&EmailPartitioningContext.bcc_addresseszsYxx& ((.9>? &&t,???)A3cHURRURS9$)zThe message part containing the actual textual email message. This is as opposed to attachments or "related" parts like an image that appears in the message etc. )preferencelist)rBget_bodycontent_type_preferencer8s r* body_part"EmailPartitioningContext.body_parts"xx  0L0L MMr,cURRS5nU(dg[RR U5nUVs/sH"n[RR U5PM$ sn$s snf)z/The "carbon-copy" Cc: addresses of the message.CcNrA)r8ccsrIrJs r* cc_addresses%EmailPartitioningContext.cc_addressesYhht$ ((-9>? &&t,???rMc*URS:XaS$S$)zWhether to prefer HTML or plain-text body when message-body has both. The default order of preference is `("html", "plain")`. The order can be switched by specifying `"text/plain"` as the `content_source` arg value. r)plainhtml)r]r\)r3rRs r*rQ0EmailPartitioningContext.content_type_preferences%)$8$8L$H _N__r,c [URURURUR(a UR/OSUR UR S9$)zThe email-specific metadata fields for this message. Suitable for use with `.metadata.update()` on the base metadata applied to message body elements by delegate partitioners for text and HTML. N) bcc_recipient cc_recipientemail_message_id sent_fromsent_tosubject)rrKrX message_id from_address to_addressesrerRs r*email_metadata'EmailPartitioningContext.email_metadatasP,,**!__-1->->t(()D%%LL   r,cURRS5nU(dg[RR U5nUVs/sH"n[RR U5PM$ nnUS$s snf)z"The address of the message sender.FromNrrA)r8fromsrIrJformatted_addrss r*rg%EmailPartitioningContext.from_addresssf  ( ((/DIJED5;;11$7EJq!!Ks)A7cURRS5nU(dgUR5RS5$)z2The value of the Message-ID: header, when present.z Message-IDNz<>)rBgetstrip)r8raw_ids r*rf#EmailPartitioningContext.message_ids3l+||~##D))r,cRUR=(d UR=(d S$)aThe best available file-path information for this email message. It's value is computed according to these rules, applied in order: - The `metadata_filename` arg value when one was provided to `partition_email()`. - The `file_path` value when one was provided. - None otherwise. This value is used as the `filename` metadata value for elements produced by partitioning the email message (but not those from its attachments). N)r4r1rRs r*r +EmailPartitioningContext.metadata_file_paths''B4??BdBr,cfUR=(d UR=(d UR$)a5The best available last-modified date for this message, as an ISO8601 string. It's value is computed according to these rules, applied in order: - The `metadata_last_modified` arg value when one was provided to `partition_email()`. - The date-time in the `Sent:` header of the message, when present. - The last-modified date recorded on the filesystem for `file_path` when it was provided. - None otherwise. This value is used as the `last_modified` metadata value for all elements produced by partitioning this email message, including any attachments. )r5 _sent_date_filesystem_last_modifiedrRs r*r/EmailPartitioningContext.metadata_last_modifieds$++`t`$B`B``r,c URb[[URS5n[[[R "U[R RS95sSSS5 $URceURR5n[[[R"U[R RS95$!,(df  Nr=f)zOThe Python stdlib `email.message.EmailMessage` object parsed from the EML file.Nrb)policy) r1openr rrDmessage_from_binary_filer}defaultr2readmessage_from_bytes)r8f file_bytess r*rBEmailPartitioningContext.msgs ?? &doot, %"@"@5<> >RR""Iz:  sA11BBcUR(dUR(d [S5eUR(aT[URR S5[ 5(d [S5eURR S5 UR[;a'[[UR5S[35eU$)z5Raise on first invalid option, return self otherwise.zOno document specified; either a `filename` or `file` argument must be provided.rz)file object must be opened in binary modez: is not a valid value for content_source; must be one of: ) r1r2r isinstancerbytesseekr3rreprrRs r*r<"EmailPartitioningContext._validate$stzza  :: "E!!LMM JJOOA    '< <,,-./$$9#:<   r,)r3r2r1r7r4r5r6)r str | NonerIO[bytes] | Nonerstrr rrrrboolr!dict[str, Any])rrrrrrr rrrrrr!rreturnr$)rzlist[str] | None)rzMIMEPart | None)rztuple[str, ...])rrrr)rr)rr)rr)rr$)__name__ __module__ __qualname____firstlineno____doc__r9 classmethodr%rrKrSrXrQrirgrfr rrBrrrerhryrxr<__static_attributes__r,r*r$r$Os `!%!%))--1$)!#  '  !+ "$  '  !+ " "*@@NN@@``    ""** C C a a e e))@@TT S Sr,r$cJ\rSrSrSrS Sjr\S Sj5rS SjrS Sjr Sr g) r'i;z8Encapsulates the partitioning logic for email documents.cXlgr0_ctx)r8r#s r*r9_EmailPartitioner.__init__>s r,c*U"US9R5$)z@Generate the document elements for the email described by `ctx`.r"_iter_elements)r=r#s r*r(_EmailPartitioner.iter_elementsAss|**,,r,c## UR5H6nURRURR5 Uv M8 URR (dgURR R5H*n[RX R5ShvN M, gN 7f)zGenerate the document elements for the email described in the partitioning context. This optionally includes elements generated by partitioning any partitionable attachments in the message as well. N) _iter_email_body_elementsmetadataupdaterrirrBiter_attachments_AttachmentPartitionerr()r8e attachments r*r _EmailPartitioner._iter_elementsFs //1A JJ  dii66 7G2yy,, ))--88:J-;;J R R R; RsB0B>2B<3 B>c#L# URRnUcgUR5nUR5n[ U[ 5(deUS:Xac[ SUURR[RURRS.URRD6ShvN g[SUURR[RURRS.URRD6ShvN gNhN7f)z/Generate document elements from the email body.Nr)textrmetadata_file_typerr) rrSget_content_type get_contentrrrr rEMLrrr)r8rS content_typecontents r*r+_EmailPartitioner._iter_email_body_elementsVsII''     113 '')'3'''' ; &%"&))">">#+<<'+yy'G'G  ))//   &"&))">">#+<<'+yy'G'G  ))//     s%B5D$7D 8A"D$D"D$"D$rN)r#r$)r#r$rIterator[Element]rr) rrrrrr9rr(rrrrr,r*r'r';s)B--S r,r'c\rSrSrSrS Sjr\S Sj5rS Sjr\ SSj5r \ SSj5r \ SSj5r S r g )ritz'Partitions an attachment to a MSG file.cXlX lgr0 _attachmentr)r8rr#s r*r9_AttachmentPartitioner.__init__ws % r,c.U"X5R5$)zHPartition an attachment MIME-part from a MIME email message (.eml file).r)r=rr#s r*r($_AttachmentPartitioner.iter_elements{s :#2244r,c## SSKJn [R"UR5nU"SUUR UR RS.UR RD6nUH"nUR UR"lUv M$ g![aZn[U[5(de[R"SUR [U5RSU35 SnAgSnAff=f7f)zPartition the byte-stream in the attachment MIME-part into elements. Generates zero elements if the attachment is not partitionable. r) partition)rrrzSkipping attachment %s: %sz: Nr)unstructured.partition.autorioBytesIO _file_bytes_attachment_file_namerrr BaseExceptionrrrwarningtyper_attached_to_filenamerattached_to_filename)r8rrelementsrs r*r%_AttachmentPartitioner._iter_elementss :zz$**+  "&"<"<'+yy'G'G))// H A.2.H.HAJJ +G a!;<< NN,**7##$Bqc*   s/'C8>B()C8 C5AC0+C80C55C8ctURRnUcg[RR U5$)z@The file-name (no path) of the message. `None` if not available.N)rr ospathbasename)r8rs r*r,_AttachmentPartitioner._attached_to_filenames1II00  ww **r,c6URR5$)zOThe original name of the attached file, `None` if not present in the MIME part.)r get_filenamerRs r*r,_AttachmentPartitioner._attachment_file_names,,..r,cURR5n[U[5(aUR S5$[U[ 5(deU$)zThe bytes of the attached file.zutf-8)rrrrencoder)r8rs r*r"_AttachmentPartitioner._file_bytessL""..0 gs # #>>'* *'5))))r,rN)rrr#r$)rrr#r$rrrr)rr)rrrrrr9rr(rrrrrrrr,r*rrtsx15%5,D5 55 B++//r,rr0)r)rrrrrrrrrrrr!rrz list[Element]),r __future__rdatetimerrD email.policy email.utilsrr email.messagerr functoolsrtypingrrr r r dateutilr unstructured.documents.elementsr runstructured.file_utils.modelrunstructured.loggerrunstructured.partition.commonr&unstructured.partition.common.metadatarunstructured.partition.htmlrunstructured.partition.textrr__annotations__r+r$r'rrr,r*rs #  0%11D2&DI660K-K -:"%$()- $-:-: -: -: " -: ' -:-:-:-:`iiX66rEEr,