j? ,UddlmZddlZddlZddlZddlZddlZddlmZddlm Z m Z m Z m Z m Z mZmZmZmZmZddlZddlmZddlmZddlmZddlmZdd lmZmZmZdd l m!Z"dd l m#Z#m$Z$dd l%m&Z&dd l'm(Z(ddl)m*Z*ddl+m,Z,ddl-m-Z-ddl.m/Z/ddl0m1Z1ddl2m3Z3ddl4m5Z5m6Z6m7Z7m8Z8m9Z9m:Z:m;Z;mZ>m!Z!m?Z?m@Z@mAZAddlBmCZCmDZDddlEmFZFmGZGmHZHddlImJZJddlKmLZLmMZMmNZNmOZOmPZPddlQmRZRmSZSmTZTeRdrddlUZUdZVdeWd<eeefZXdeWd<ee*e"fZYdeWd <eTdddd!d!ddd"gd#fd;d3ZZeAeDeCj[e1dddd!d!d!ddd"gd#f d open(filename, "rb"). include_metadata Determines whether or not metadata is included in the metadata attribute on the elements in the output. infer_table_structure If True, any Table elements that are extracted will also have a metadata field named "text_as_html" where the table's text content is rendered into an html string. I.e., rows and cells are preserved. Whether True or False, the "text" field is always present in any Table element and is the text content of the table (no structure). languages User defined value for `metadata.languages` if provided. Otherwise language is detected using naive Bayesian filter via `langdetect`. Multiple languages indicates text could be in either language. Additional Parameters: detect_language_per_element Detect language per element instead of at the document level. rDrFrDr>rPcjtj|std|d|S)z;Return path to a file confirmed to exist on the filesystem.z The file z does not exist.)ospathexists ValueError)rDs e/lsinfo/ai/hellotax_ai/base_platform/venv/lib/python3.11/site-packages/unstructured/partition/docx.pyvalidate_filenamez5convert_and_partition_docx..validate_filename}s8w~~h'' ECCCCDD DrF IO[bytes]ctjd5}|||jcdddS#1swxYwYdS)z6Return path to temporary copy of file to be converted.F)deleteN)tempfileNamedTemporaryFilewritereadname)rFtmps rYcopy_to_tempfilez4convert_and_partition_docx..copy_to_tempfiles  ( 6 6 6 # IIdiikk " " "8                  s.AAA file_pathctj|}tj|\}}|dS)zAReturn a filename like "foo.docx" from a path like "a/b/foo.odt" z.docx)rUrVbasenamesplitext)rfrD root_name_s rYextract_docx_filenamez9convert_and_partition_docx..extract_docx_filenamesA7##I..w''11 1""""r[r=)format outputfile)rDrKrHrJrLrMrON)rDr>rPr>)rFr\rPr>)rfr>rPr>) r0rrbytesr_TemporaryDirectoryrUrVjoinr< convert_filepartition_docx)rCrDrFrHrJrKrLrMrOrZrerlrftmpdir docx_pathelementss rYconvert_and_partition_docxrwRsiR----  ####08d!!(+++=M=MdSUV[S\^bNcNc=d=dI  $ & & &GLL)>)>y)I)IJJ         "/-"7#9(C                  $ Os'AC  CCinclude_page_breakschunking_strategykwargsrc  t||t||||||} t| || } t | S)a.Partitions Microsoft Word Documents in .docx format into its document elements. Parameters ---------- filename A string defining the target filename path. file A file-like object using "rb" mode --> open(filename, "rb"). infer_table_structure If True, any Table elements that are extracted will also have a metadata field named "text_as_html" where the table's text content is rendered into an html string. I.e., rows and cells are preserved. Whether True or False, the "text" field is always present in any Table element and is the text content of the table (no structure). metadata_filename The filename to use for the metadata. Relevant because partition_doc converts the document to .docx before partition. We want the original source filename in the metadata. metadata_last_modified The last modified date for the document. languages User defined value for `metadata.languages` if provided. Otherwise language is detected using naive Bayesian filter via `langdetect`. Multiple languages indicates text could be in either language. Additional Parameters: detect_language_per_element Detect language per element instead of at the document level. rS)rvrMrO)r0_DocxPartitioneriter_document_elementsr3list) rDrFrKrxrHrJrLryrMrOrzrvs rYrsrsslX----66  H#$?H >>r[ceZdZdZ dIdJdZe dIdKdZdLdZdLdZdMdZ dNdOdZ e dPdZ e dQdZ e dQd ZdRd#ZdSd%ZdTd&ZdMd'ZdUd)ZdVd-ZdWd/ZdXd2ZdYd4ZdZd5Zd[d7Ze d\d8Zed]d:Zd^d<Zd_d>Zd`d@ZdadAZ dbdBZ!dcdDZ"dddFZ#dddGZ$dedHZ%dS)fr|z8Provides `.partition()` for MS-Word 2007+ (.docx) files.NTrDrErFrGrKrxrIrJrLrPNonech||_||_||_||_||_||_d|_dS)N) _filename_file_metadata_filename_include_page_breaks_infer_table_structure_metadata_last_modified _page_counter)selfrDrFrKrxrJrLs rY__init__z_DocxPartitioner.__init__s@" "3$7!&;#'=$"#r[Iterator[Element]c|||||||}|jr|n|S)zFPartition MS Word documents (.docx format) into its document elements.)rDrFrKrxrJrL)_document_contains_sections_iter_document_elements#_iter_sectionless_document_elements)clsrDrFrKrxrJrLrs rYr}z'_DocxPartitioner.iter_document_elementss^s/ 3"7#9    / t |}|rt ||t VdS||}|r|||t VdS||}|r|||t VdSt||t VdS)aGenerate zero-or-one document element for `paragraph`. In Word, an empty paragraph is commonly used for inter-paragraph spacing. An empty paragraph does not contribute to the document-element stream and will not cause an element to be emitted. N)textmetadatadetection_origin)rr) rstrip_paragraph_metadata _is_list_itemr r(r?_style_based_element_type&_parse_paragraph_text_for_element_typer+)rrrr clean_text TextSubClss rY_classify_paragraph_to_elementz/_DocxPartitioner._classify_paragraph_to_elementFsG~zz||  F++I66   i ( ( &t,,2244J #%%5 F33I>>  *$L\]]] ] ] ] F@@KK  *$L\]]] ] ] ] F4(=MNNNNNNNNr[Ftabler is_nestedr>cnd fd dfd tfd |jD|rgnd d S)aHTML string version of `table`. Example:
item qty
spam 42
eggs 451
bacon 0
`is_nested` is used for recursive calls when a nested table is encountered. Certain behaviors are different in that case, but the caller can safely ignore that parameter and allow it to take its default value. cellrrP Iterator[str]c3K|D]b}t|trtj|jV4t|t r|dVcdS)NT)r)rrrhtmlescaperr_convert_table_to_htmlrrrs rYiter_cell_block_itemszF_DocxPartitioner._convert_table_to_html..iter_cell_block_itemss"5577 R R j)44R"[99;;;;; R55jD5QQQQQ R Rr[rowrc*fd|jDS)Nc3TK|]"}d|V#dS) Nrq).0rrs rY zN_DocxPartitioner._convert_table_to_html..iter_cells..s;QQtDII33D99::QQQQQQr[)cells)rrs rY iter_cellsz;_DocxPartitioner._convert_table_to_html..iter_cellssQQQQsyQQQ Qr[c@g|]}t|S)r~)rrrs rY z;_DocxPartitioner._convert_table_to_html..s) 9 9 9sT**S// " " 9 9 9r[firstrow unsafehtml)headerstablefmtrrrPr)rrrPr)rrows)rrrrrs` @@rYrz'_DocxPartitioner._convert_table_to_htmlqs& R R R R R R R R R R R R 9 9 9 9ej 9 9 9#3BB"     r[rc|j|j}}|tj|S|Jt |t r;|dtj| }tj|S)z?The python-docx `Document` object loaded from file or filename.Nr) rrr=rrrseekioBytesIOrb)rrDrFs rYrz_DocxPartitioner._documents|$  =** * d0 1 1 + IIaLLL:diikk**D}T"""r[c^d}t|jj|S)a>True when there is at least one page-break detected in the document. Only `w:lastRenderedPageBreak` elements reliably indicate a page-break. These are reliably inserted by Microsoft Word, but probably don't appear in documents converted into .docx format from for example .odt format. z./w:body/w:p/w:r/w:lastRenderedPageBreak | ./w:body/w:p/w:hyperlink/w:r/w:lastRenderedPageBreak | ./w:body/w:tbl/w:tr/w:tc/w:p/w:r/w:lastRenderedPageBreak | ./w:body/w:tbl/w:tr/w:tc/w:p/w:hyperlink/w:r/w:lastRenderedPageBreak)rIrelementxpath)rrs rY_document_contains_pagebreaksz._DocxPartitioner._document_contains_pagebreakss0 V DN*0077888r[c4t|jjS)a2True when there is at least one section in the document. This is always true for a document produced by Word, but may not always be the case when the document results from conversion or export. In particular, a Microsoft Teams chat-transcript export will have no sections. )rIrrrs rYrz,_DocxPartitioner._document_contains_sectionssDN+,,,r[hdrftr_Header | _Footerc`dfd }dd||DS) amThe text enclosed in `hdrftr` as a single string. Each paragraph is included along with the text of each table cell. Empty text is omitted. Each paragraph text-item is separated by a newline (" ") although note that a paragraph that contains a line-break will also include a newline representing that line-break, so newlines do not necessarily distinguish separate paragraphs. The entire text of a table is included as a single string with a space separating the text of each cell. A header with no text or only whitespace returns the empty string (""). rrrPrc3K|D]r}t|tr|jV3t|t r*d|VsdS)zGenerate each text item in `hdrftr` stripped of leading and trailing whitespace. This includes paragraphs as well as table cell contents.  N)rrrrrrrq_iter_table_texts)rrrs rYiter_hdrftr_textsz?_DocxPartitioner._header_footer_text..iter_hdrftr_textss %7799 G G j)44G$///111111 G((4#9#9*#E#EFFFFF G Gr[rc3K|]}||V dSNrrrs rYrz7_DocxPartitioner._header_footer_text..s'LL$tLLLLLLLr[)rrrPrr)rrrs` rY_header_footer_textz$_DocxPartitioner._header_footer_textsV G G G G G GyyLL*;*;F*C*CLLLLLLr[Iterator[PageBreak]c#lK|xjdz c_|jrtdtVdSdS)zGIncrement page-number by 1 and generate a PageBreak element if enabled.r)rN)rrr*r?rs rY_increment_page_numberz'_DocxPartitioner._increment_page_numbersR a  $ CB1ABBB B B B B B C Cr[cJt|jrdSd|jjvS)z7True when `paragraph` can be identified as a list-item.Tz )r4r_pxml)rrs rYrz_DocxPartitioner._is_list_items) IN + + 4il...r[c#Kdfd |D]M}t|tr||Ed{V3|Ed{VNdS)zGenerate zero-or-more document elements for `paragraph`. The generated elements can be both textual elements and PageBreak elements. An empty paragraph produces no elements. rrrP'Iterator[Paragraph | RenderedPageBreak]c3K|js|VdS|jd}|j}|r|V|V|j}|r|Ed{VdSdS)anGenerate Paragraph and RenderedPageBreak items from `paragraph`. Each generated paragraph is the portion of the paragraph on the same page. When the paragraph contains no page-breaks, it is iterated unchanged and iteration stops. When there is a page-break, in general there one paragraph "fragment" before the page break, the page break, and then the fragment after the page break. However many combinations are possible. The first item can be either a page-break or a paragraph, but the type always alternates throughout the sequence. Nr)contains_page_breakrendered_page_breakspreceding_paragraph_fragmentfollowing_paragraph_fragment)r page_breakrriter_paragraph_itemss rYrzG_DocxPartitioner._iter_paragraph_elements..iter_paragraph_itemss0 "7:J,6+R (+ 32222    ,6+R (, N//0LMMMMMMMMMMM N Nr[N)rrrPr)rrrr)rritemrs @rYrz)_DocxPartitioner._iter_paragraph_elementss N N N N N NB)(33 9 9D$ ** 9>>tDDDDDDDDDD668888888888  9 9r[Iterator[Dict[str, str]]c#K|jD]C}|jr|jnd}|s'|jr|ddV|jr|ddVDdS)zLGenerate e.g. {"text": "MUST", "tag": "b"} for each emphasis in `paragraph`.rb)rtagiN)runsrrbolditalic)rrrunrs rY_iter_paragraph_emphasisz)_DocxPartitioner._iter_paragraph_emphasis!s> 1 1C'*x738>>###RD x 1#C00000z 1#C00000 1 1r[rrIterator[Footer]c#Kd fd }||jdEd {V|jr||jd Ed {Vjjjr||jd Ed {Vd Sd S) aOGenerate any `Footer` elements defined for this section. A Word document has up to three header and footer definition pairs for each document section, a primary, first-page, and even-page header and footer. The first-page pair applies only to the first page of the section (perhaps a title page or chapter start). The even-page pair is used in book-bound documents where there are both recto and verso pages (it is applied to verso (even-numbered) pages). A page where neither more specialized footer applies uses the primary footer. footerrheader_footer_typer>rPrc 3K|jrdS|}|sdSt|tt j|dVdS)z2Generate zero-or-one Footer elements for `footer`.NrrDrcategory_depthrrr)is_linked_to_previousrr%r?r#r)rrrrs rY iter_footerz;_DocxPartitioner._iter_section_footers..iter_footer7+ ++F33D !1(!4'9#$     r[primaryN first_page even_page)rrrr>rPr)r"different_first_page_header_footerfirst_page_footerrsettings odd_and_even_pages_header_footereven_page_footer)rrrs` rYrz&_DocxPartitioner._iter_section_footers,s      ";w~y999999999  5 L"{7# " C J"{7#;[II I I I I I I I I I J Jr[Iterator[Header]c#Kd fd }||jdEd {V|jr||jd Ed {Vjjjr||jd Ed {Vd Sd S) zGenerate `Header` elements for this section if it has them. See `._iter_section_footers()` docstring for more on docx headers and footers. headerrrr>rPrc 3K|jrdS|}|sdSt|tt j|dVdS)z2Generate zero-or-one Header elements for `header`.Nrrr)rrr&r?r#r)rrrrs rYmaybe_iter_headerzA_DocxPartitioner._iter_section_headers..maybe_iter_headerTrr[rNrr )rrrr>rPr)rr first_page_headerrr r even_page_header)rrrs` rYrz&_DocxPartitioner._iter_section_headersNs       "%$W^Y?????????  5 R(()BLQQ Q Q Q Q Q Q Q > " C P(()A;OO O O O O O O O O O P Pr[rintc#Kdfd }|j}|tjkr%|sEd{Vn<|tjkr,|dkrdS|rEd{VdS)aGenerate zero-or-one `PageBreak` document elements for `section`. A docx section has a "start" type which can be "continuous" (no page-break), "nextPage", "evenPage", or "oddPage". For the next, even, and odd varieties, a `w:renderedPageBreak` element signals one page break. Here we only need to handle the case where we need to add another, for example to go from one odd page to another odd page and we need a total of two page-breaks. rPrIc jdzdkS)Nr)rrsrY page_is_oddz?_DocxPartitioner._iter_section_page_breaks..page_is_oddus%)Q. .r[NrrPrI) start_typer EVEN_PAGErODD_PAGE)rrrrrs` rYrz*_DocxPartitioner._iter_section_page_breaksks / / / / / /'  )3 3 3;== 966888888888 +4 4 4a{}} 966888888888 r[Iterator[Table]c #DK|jr||nd}d||}||\}}t |t t||j|j |j |pd|pdVdS)zBGenerate zero-or-one Table element for a DOCX `w:tbl` XML element.Nr) text_as_htmlrD page_number last_modifiedemphasized_text_contentsemphasized_text_tags)rr) rrrqr_table_emphasisrr?r#r _page_number_last_modified)rr html_table text_tabler$r%s rYrz$_DocxPartitioner._iter_table_elements<@;V`T00777\` XXd44U;;<< 9=9M9Me9T9T6 "6 -$'0 -"1)A)IT%9%AT     r[c#|K|jD]1}|jD]'}|jD]}||Ed{V(2dS)zHGenerate e.g. {"text": "word", "tag": "b"} for each emphasis in `table`.N)rr paragraphsr)rrrrrs rY_iter_table_emphasisz%_DocxPartitioner._iter_table_emphasiss: H HC  H H!%HHI#<.iter_cell_textss #5577 B B j)44B$///111111 B $55jAAAAAAAAA B Br[continuec3K|]}||V dSrrrs rYrz5_DocxPartitioner._iter_table_texts..s(UUTPTUDUUUUUUr[Nr)r_trtc_lstvMerger)rrr0rtrtcs` rYrz"_DocxPartitioner._iter_table_textss B B B B B B: V VCBi V V9 **UU__U2s^^-L-LUUUUUUUUUUU  V V Vr[c|jr|jS|j|j}}|&|drdnt |S|Jt |S)z8Last-modified date suitable for use in element metadata.Nz/tmp)rrr startswithr1r2)rrfrFs rYr(z_DocxPartitioner._last_modifiedsp  ' 0/ /.$*4   $//77^44=ST]=^=^ ^/555r[ Optional[int]c"|jr|jndS)a4The current page number, or None if we can't really tell. Page numbers are not added to element metadata if we can't find any page-breaks in the document (which may be a common case). In the DOCX format, determining page numbers is strictly a best-efforts attempt since actual page-breaks are determined at rendering time (e.g. printing) based on the fontmetrics of the target device. Explicit (hard) page-breaks are always recorded in the docx file but the rendered page-breaks are only added optionally. N)rrrs rYr'z_DocxPartitioner._page_numbers&*%GQt!!TQr[Tuple[List[str], List[str]]ctj||\}}d|Dd|DfS)z@[contents, tags] pair describing emphasized text in `paragraph`.cg|] }|d Srrres rYrz8_DocxPartitioner._paragraph_emphasis..s000q6000r[cg|] }|d Srrr@s rYrz8_DocxPartitioner._paragraph_emphasis..s2S2S2S1U82S2S2Sr[) itertoolsteer)rr iter_p_emph iter_p_emph_2s rY_paragraph_emphasisz$_DocxPartitioner._paragraph_emphasissM%.]43P3PQZ3[3[%\%\" ]00K0002S2S]2S2S2STTr['Tuple[List[str], List[str], List[Link]]cjsgggfSdfd }t|}d|D}d|D}|||fS)z,Describes hyperlinks in `paragraph`, if any.rPIterator[Link]c3:Kd}D]}t|tr|t|jz }/t|t r<|j}|j}|}|t|z }|si|slt|||VdS)zGenerate `Link` typed-dict for each external link in `paragraph`. Word uses hyperlinks for internal "jumps" within the document, as well as for web and other external locations. Only generate the external ones. r)rurl start_indexN)rrrlenrrrMr')offsetrrrMrNrs rYiter_paragraph_linkszC_DocxPartitioner._paragraph_link_meta..iter_paragraph_linkss F!4466 L LdC((Lc$)nn,FFi00L9D(C"(Kc$ii'F !  ! Dc{KKKKKK' L Lr[c"g|] }|dpd S)rrrrlinks rYrz9_DocxPartitioner._paragraph_link_meta..s!;;;Td6l(b;;;r[cg|] }|d S)rMrrSs rYrz9_DocxPartitioner._paragraph_link_meta..s333TT%[333r[)rPrK) hyperlinksr~)rrrQlinks link_texts link_urlss ` rY_paragraph_link_metaz%_DocxPartitioner._paragraph_link_metas# r2:  L L L L L L8))++,,<;U;;; 33U333 9e++r[r#c ||}||\}}||\}}}t||pd|pd|j|j|pd|pd|pd|j }d|_|S)z.ElementMetadata object describing `paragraph`.N) rr$r%rDr#rXrYrWr"r=)_parse_category_depth_by_stylerHrZr#rr(r'r) rrrr$r%rXrYrWelement_metadatas rYrz$_DocxPartitioner._paragraph_metadatas<)/*>K8 88DD 199L;;== =r[cdS)Nrrrs rYrcz4_DocxPartitioner._parse_category_depth_by_style_ilvl>sqr[rdcd d}dr |SdkrdSgd }tfd |Dr |Sd S) zParse category-depth from the style-name of `paragraph`. Category depth is 0-indexed and relative to the other element types in the document. suffixr>rPrc|dr*t|ddz ndS)Nrr)splitisdigitr)rhs rY_extract_numberzM_DocxPartitioner._parse_category_depth_by_style_name.._extract_numberHsF28,,..2D2L2L2N2NU3v||~~b)**Q..TU Ur[HeadingSubtitler)r List Bullet List Continue List Numberc3BK|]}|VdSr)r9)rprefixrds rYrzG_DocxPartitioner._parse_category_depth_by_style_name..Ts1IIz$$V,,IIIIIIr[r)rhr>rPr)r9any)rrdrm list_prefixess ` rYrbz4_DocxPartitioner._parse_category_depth_by_style_nameBs  V V V V   + + /"?:.. .  # #1POO IIII=III I I /"?:.. .qr[Optional[Type[Text]]c|j}t|dkrdSt|rtSt |rt St|rtSt|rtSdS)zEAttempt to differentiate the element-type by inspecting the raw text.rN) rrrOr8r!r5r$r6r)r7r,)rrrs rYrz7_DocxPartitioner._parse_paragraph_text_for_element_typeZs~##%% t99q==4  % % N D ! !   %d + + ! T " " Ltr[c idtdtdtdtdtdtdtdtd td td td td tdtdtdtdttttttttttttttd }|jr |jjpd}||S)zElement-type for `paragraph` based on its paragraph-style. Returns `None` when the style doesn't tell us anything useful, including when it is the default "Normal" style. Captionz Heading 1z Heading 2z Heading 3z Heading 4z Heading 5z Heading 6z Heading 7z Heading 8z Heading 9z Intense Quoter zList 2zList 3rpz List Bullet 2z List Bullet 3) rqzList Continue 2zList Continue 3rrz List Number 2z List Number 3zList Paragraphz Macro Textz No SpacingQuotero TOCHeadingr,r_)r+r,r(rarcget)rrSTYLE_TO_ELEMENT_MAPPINGrds rYrz*_DocxPartitioner._style_based_element_typeks9$ t$ $  $   $  $  $  $  $  $  $  T$  H$  h$  h$  8$ X!$ " X#$ $&''#%%&=$ $ $  F o>)/*>K8 (++J777r[ctj||\}}d|Dd|DfS)z<[contents, tags] pair describing emphasized text in `table`.cg|] }|d Sr?rr@s rYrz4_DocxPartitioner._table_emphasis..s222q6222r[cg|] }|d SrCrr@s rYrz4_DocxPartitioner._table_emphasis..s4W4W4W!QuX4W4W4Wr[)rDrEr-)rr iter_tbl_emphiter_tbl_emph_2s rYr&z _DocxPartitioner._table_emphasissM)2t7P7PQV7W7W)X)X& 22M2224W4W4W4W4WXXr[)NNNTTN)rDrErFrGrKrErxrIrJrIrLrErPr)rDrErFrGrKrErxrIrJrIrLrErPr)rPr)rrrPr)F)rrrrIrPr>)rPrr)rrrPr>)rPr)rrrPrI)rrrPr)rrrPr)rrrPr)rrrrrPr)rrrPr)rrrPr)rrrPr)rPrE)rPr:)rrrPr<)rrrPrI)rrrPr#)rrrPr)rPr)rdr>rPr)rrrPrw)rrrPr<)&__name__ __module__ __qualname____doc__r classmethodr}rrrrr:rrrrrrrrrrrrr-rr(propertyr'rHrZrr\rcrbrrr&rr[rYr|r|s:BB#'$(+/$(&*04$$$$$*#'$(+/$(&*04    [ 8;;;;: @ @ @ @)O)O)O)OV( ( ( ( ( T # # #\ #999\9(---\-MMMM<CCCC ////,9,9,9,9\ 1 1 1 1 J J J JDPPPP:$$$$L    *HHHHVVVV<666\6$ R R RX RUUUU &,&,&,&,P    &>>>>$0"08080808dYYYYYYr[r|)rCr>rDrErFrGrHrIrJrIrKrErLrErMrNrOrIrPrQ)rDrErFrGrKrErxrIrHrIrJrIrLrEryrErMrNrOrIrzrrPrQ)^ __future__rrrrDrUr_rtypingrrrrr r r r r rr= docx.documentrdocx.enum.sectionrdocx.oxml.tablerdocx.oxml.text.paragraphr docx.sectionrrr docx.tablerrrrdocx.text.hyperlinkrdocx.text.pagebreakrdocx.text.paragraphr docx.text.runrrtyping_extensionsrunstructured.chunkingrunstructured.cleaners.corer unstructured.documents.elementsr!r"r#r$r%r&r'r(r)r*r+r,r- unstructured.file_utils.filetyper.r/unstructured.partition.commonr0r1r2unstructured.partition.langr3 unstructured.partition.text_typer4r5r6r7r8unstructured.utilsr9r:r;r<r?__annotations__r@rArwDOCXrsr|rr[rYrs#""""""  ))))))                         """"""......""""""))))))2222222222))))))""""""""))))))111111))))))''''''777777444444 RQQQQQQQ <;;;;;VUUUUUUUUUZ  OOOf - ----Y 12 2222z""# $!"&'+,0&,X(-SSSS#"SlHM**" $'+ $!"&,0'+&,X(-8888+*8vy Yy Yy Yy Yy Yy Yy Yy Yy Yy Yr[