B ©‰Ad6|ã@sdZddlmZddlmZmZmZmZmZm Z ddl m Z ddl Z ddl Z ddlmZddlmZdd lmZdd lmZddlZdd lmZdd lmZmZmZdd lmZyddl m!Z!Wn e"k ràddl#m!Z!YnXddl$Z$ddl%Z%ddl&Z&ddlZddl'Z'Gdd„de(ƒZ)dS)z/ Created : 2015-03-12 @author: Eric Lapouyade é)ÚPathLike)ÚAnyÚOptionalÚIOÚUnionÚDictÚSeté)ÚSubdocN)Úetree)ÚDocument)Ú parse_xml)ÚXmlPart)ÚRELATIONSHIP_TYPE)Ú EnvironmentÚTemplateÚmeta)Ú TemplateError)Úescapec@sÎeZdZdZdZdZeeee e fddœdd„Z dXe d œd d „Z d d „Zdd„ZdYdd„Zdd„Zdd„Zdd„Zdd„ZdZdd„Zd[ee efeeddœdd„Zd d!„Zd\d"d#„Zd$d%„Zd&d'„Zd(d)„Zd*d+„Z d]d,d-„Z!d.d/„Z"d^ee efeee dd1œd2d3„Z#d4d5„Z$d6d7„Z%d_d8d9„Z&e'd:d;„ƒZ(dd?„Z*d@dA„Z+dBdC„Z,dDdE„Z-dFdG„Z.dHdI„Z/dJdK„Z0dLdM„Z1dNdO„Z2dPdQ„Z3eeee e fddRœdSdT„Z4d`eee5e dUœdVdW„Z6e7e6ƒZ8dS)aÚ DocxTemplatez= Class for managing docx files as they were jinja2 templates zJhttp://schemas.openxmlformats.org/officeDocument/2006/relationships/headerzJhttp://schemas.openxmlformats.org/officeDocument/2006/relationships/footerN)Ú template_fileÚreturncCs$||_| ¡d|_d|_d|_dS)NF)rÚreset_replacementsÚdocxÚ is_renderedÚis_saved)Úselfr©rúB/var/www/html/venv/lib/python3.7/site-packages/docxtpl/template.pyÚ__init__'s zDocxTemplate.__init__T)ÚreloadcCs&|jr|jr"|r"t|jƒ|_d|_dS)NF)rrr r)rr rrrÚ init_docx.s zDocxTemplate.init_docxcCs$| ¡i|_d|_d|_d|_dS)NièF)r!Úpic_mapÚcurrent_rendering_partÚdocx_ids_indexr)rrrrÚ render_init3s zDocxTemplate.render_initcCs t|j|ƒS)N)Úgetattrr)rÚnamerrrÚ __getattr__:szDocxTemplate.__getattr__ÚunicodecCstj|dddS)Nr)F)ÚencodingZ pretty_print)r Útostring)rÚxmlr*rrrÚ xml_to_string=szDocxTemplate.xml_to_stringcCs| ¡|jS)N)r!r)rrrrÚget_docxBszDocxTemplate.get_docxcCs| |jjj¡S)N)r-rÚ_elementÚbody)rrrrÚget_xmlFszDocxTemplate.get_xmlc Cs(t|dƒ}| | ¡¡WdQRXdS)NÚw)ÚopenÚwriter1)rÚfilenameÚfhrrrÚ write_xmlIs zDocxTemplate.write_xmlc Csftjdd|tjd}dd„}tjd||tjd}dd„}tjd ||tjd}d d „}tjd ||tjd}tjd d|tjd}tjdd|tjd}tjdd|tjd}tjdd|tjd}x,dD]$}dd|i}tj|d|tjd}q¾Wx,dD]$}dd|i}tj|d|tjd}qìWdd„}tjd||tjd}dd„}tjd ||tjd}d!d"„} t d#| |¡}|S)$zÅ Make a lots of cleaning to have a raw xml understandable by jinja2 : strip all unnecessary xml tags, manage table cell background color and colspan, unescape html entities, etc... z8(?<={)(<[^>]*>)+(?=[\{%\#])|(?<=[%\}\#])(<[^>]*>)+(?=\})Ú)ÚflagscSstjdd| d¡tjdS)Nz.*?(|]*>)r8r)r9)ÚreÚsubÚgroupÚDOTALL)ÚmrrrÚ striptagsYsz)DocxTemplate.patch_xml..striptagsz,{%(?:(?!%}).)*|{#(?:(?!#}).)*|{{(?:(?!}}).)*cSsR| d¡| d¡}tjdd|tjd}tjdd|dd}t dd | d ¡|¡S) Nr éz.](?:(?!]).)*.*?r8)r9z)Úcountz(]*>)z\1é)r<r:r;r=)r>Úcell_xmlrrrÚcolspan`s z'DocxTemplate.patch_xml..colspanzE(](?:(?!]).)*){%\s*colspan\s+([^%]*)\s*%}(.*?)cSsR| d¡| d¡}tjdd|tjd}tjdd|dd}t dd | d ¡|¡S) Nr r@z.](?:(?!]).)*.*?r8)r9z )rAz(]*>)z7\1rB)r<r:r;r=)r>rCrrrÚcellbgksz&DocxTemplate.patch_xml..cellbgzD(](?:(?!]).)*){%\s*cellbg\s+([^%]*)\s*%}(.*?)z'((?:(?!).)*)({{.*?}}|{%.*?%})z\1\2z({{r\s.*?}}|{%r\s.*?%})zX\1z(?:(?!).)*?{%-z{%z'-%}(?:(?!]|{%|{{).)*?]*?>z%})ÚtrÚtcÚpÚrzP](?:(?!]).)*({%%|{{)%(y)s ([^}%%]*(?:%%}|}})).*?Úyz\1 \2)rFrGrHzG](?:(?!]).)*({#)%(y)s ([^}#]*(?:#})).*?cSs dd„}tjd|| ¡tjdS)NcSs4d| d¡d| d¡| d¡d| d¡S)NzKr z{% if loop.first %}rBr@z {% endif %}é)r<)Úm1rrrÚv_merge”s(z;DocxTemplate.patch_xml..v_merge_tc..v_mergez?(].*?)(.*?)(?:{%\s*vm\s*%})(.*?)())r9)r:r;r<r=)r>rMrrrÚ v_merge_tc“s  z*DocxTemplate.patch_xml..v_merge_tcz6](?:(?!]).)*?{%\s*vm\s*%}.*?]cSsn| ¡}dd„}dd„}t d|¡rNtjd||tjd}tjdd |tjd}ntjd ||tjd}d |d S) NcSs&| d¡d| d¡d| d¡S)Nr z{{ rBz * loop.length }}r@)r<)rLrrrÚ with_gridspan¬szADocxTemplate.patch_xml..h_merge_tc..with_gridspancSs,d| d¡| d¡| d¡| d¡S)Nz'r rBr@rK)r<)Úm2rrrÚwithout_gridspan³s zDDocxTemplate.patch_xml..h_merge_tc..without_gridspanz w:gridSpanz(w:gridSpan w:val=")(\d+)("))r9z {%\s*hm\s*%}r8z?(].*?)(.*?)(?:{%\s*hm\s*%})(.*?)()z{% if loop.first %}z {% endif %})r<r:Úsearchr;r=)r>Z xml_to_patchrOrQr,rrrÚ h_merge_tc©s(    z*DocxTemplate.patch_xml..h_merge_tcz6](?:(?!]).)*?{%\s*hm\s*%}.*?]cSsB| d¡ dd¡ dd¡ dd¡ dd ¡ d d ¡ d d¡ d d¡S) Nrz‘ú'z<úu“ú"uâ€u‘u’)r<Úreplace)r>rrrÚ clean_tagsÚs z*DocxTemplate.patch_xml..clean_tagsz(?<=\{[\{%])(.*?)(?=[\}%]}))r:r;r=) rÚsrc_xmlr?rDrErJÚpatrNrSrYrrrÚ patch_xmlMsD        . zDocxTemplate.patch_xmlc Csæt dd|¡}y,||_|r&| |¡}nt|ƒ}| |¡}Wnjtk r¤}zLt|dƒr|jdk rt |jddƒ}t dd„|  ¡||d…ƒ|_ |‚Wdd}~XYnXt d d |¡}|  d d ¡  d d¡  dd¡  dd¡}| |¡}|S)Nz ])z\n]+>r8)r:r;)ÚxrrrÚóóz.DocxTemplate.render_xml_part..éz \n])z|D]6}t|jj|ƒ}| |¡}| |¡}t|jj||ƒq$WdS)NÚauthorÚcommentsÚ identifierÚlanguageÚsubjectÚtitle)rr&rZcore_propertiesrbrcÚsetattr)rrjrkÚ propertiesÚpropÚinitialrlÚrenderedrrrÚrender_propertiesÿs   zDocxTemplate.render_propertiescs8dd„‰‡fdd„‰‡fdd„}tjd||tjd}|S) NcSsR| d¡ dd||f¡}| dd||f¡}| dd¡}| dd ||f¡}|S) Nrú zB%s%súz:%s%sÚ z'ú ze%s%s)r<rX)Úrun_propertiesÚparagraph_propertiesr>r,rrrÚ resolve_texts    z2DocxTemplate.resolve_listing..resolve_textcsJt d| d¡¡‰ˆr ˆ d¡nd‰tjd‡‡‡fdd„| d¡tjdS)Nz.*?rr8z]*)?>.*?cs ˆˆˆ|ƒS)Nr)r^)rr‚r€rrr_*r`zCDocxTemplate.resolve_listing..resolve_run..)r9)r:rRr<r;r=)rr>)r‚)rr€rÚ resolve_run&s z1DocxTemplate.resolve_listing..resolve_runcsHt d| d¡¡‰ˆr ˆ d¡nd‰tjd‡‡fdd„| d¡tjdS)Nz.*?rr8z]*)?>.*?cs ˆˆ|ƒS)Nr)r^)rrƒrrr_1r`zIDocxTemplate.resolve_listing..resolve_paragraph..)r9)r:rRr<r;r=)r>)rƒ)rrÚresolve_paragraph-s  z7DocxTemplate.resolve_listing..resolve_paragraphz]*)?>.*?)r9)r:r;r=)rr,r„r)rƒr‚rrhs   zDocxTemplate.resolve_listingcCs*| ¡}| |¡}| ||jj||¡}|S)N)r1r\rorÚ_part)rrjrkr,rrrÚ build_xml8s zDocxTemplate.build_xmlcCs|jj}|j}| ||¡dS)N)rr/r0rX)rÚtreeÚrootr0rrrÚmap_tree>szDocxTemplate.map_treeccs>x8|jjj ¡D]&\}}|j|kr|jjr||jfVqWdS)N)rr…Ú_relsÚitemsÚreltypeÚ target_partÚblob)rÚuriÚrelKeyÚvalrrrÚget_headers_footersCsz DocxTemplate.get_headers_footerscCs| t|jƒ¡S)N)r-r rŽ)rrirrrÚ get_part_xmlHszDocxTemplate.get_part_xmlcCs"t d|tj¡}|r| d¡SdS)Nz <\?xml[^\?]+\bencoding="([^"]+)"r zutf-8)r:ÚmatchÚIr<)rr,r>rrrÚget_headers_footers_encodingKs z)DocxTemplate.get_headers_footers_encodingccsZxT| |¡D]F\}}| |¡}| |¡}| |¡}| ||||¡}|| |¡fVq WdS)N)r’r“r–r\roÚencode)rrjrrkrrir,r*rrrÚbuild_headers_footers_xmlQs    z&DocxTemplate.build_headers_footers_xmlcCsj|jjj|j}t |j|j||j¡}x.|j   ¡D] \}}|  |j |j |j|j¡q2W||jjj|_ dS)N)rr…rŠrrÚloadZpartnameÚ content_typeÚpackageÚrelsr‹Zload_relrŒÚ_targetÚrIdÚ is_external)rrr,riZnew_partržÚrelrrrÚmap_headers_footers_xmlYs z$DocxTemplate.map_headers_footers_xmlF)rjrkÚ autoescaperc Cs¾| ¡|r"|st|d}n||_| ||¡}| |¡}| |¡| |¡| ||j|¡}x|D]\}}|  ||¡qbW| ||j |¡} x| D]\}}|  ||¡qW|  ||¡d|_ dS)N)r¢T) r%rr¢r†Ú fix_tablesÚ fix_docpr_idsr‰r˜Ú HEADER_URIr¡Ú FOOTER_URIr{r) rrjrkr¢Zxml_srcr‡Úheadersrr,Zfootersrrrrc`s&        zDocxTemplate.renderc sºtjdd}tj||d}d|jdd‰x†| ˆd¡D]r}| ˆd¡}| ˆd ¡}d }xH| ˆd ¡D]6}| ˆd ¡} t|ƒ|t| ƒkrtt| ƒt|ƒ}qtW|d krd } d} x4|D],} |  ˆd¡dk rÆ| t |  ˆd¡ƒ7} qÆW| d kr| t|ƒ} | t|ƒ|} x<|D]4} |   ˆdt t t |  ˆd¡ƒ| | ƒƒ¡q"Wx4t |ƒD](}t |ˆd ˆdt t | ƒƒi¡qdW| ˆd ¡}t|ƒ}d }‡fdd„}x>| ˆd ¡D],}| ˆd ¡} t || d ¡}t||ƒ}qÆW||}|d kr>d }x8|| d…D]&} |t |  ˆd¡ƒ7}| | ¡qW| ˆd ¡}d }t|ƒd krx|t|ƒ}t |ƒ}x8|D]0} |   ˆdt t t |  ˆd¡ƒ|ƒƒ¡q~Wq>W|S)NT)Zrecover)ÚparserÚ{r2Ú}ÚtblÚtblGridZgridColrrFrGgcsN| ˆd¡}|dkrdn | ˆd¡}|dk rF|t| ˆd¡ƒS|dS)NZtcPrZgridSpanr‘r )ÚfindÚintÚget)ÚtotalÚcellZtc_prZ grid_span)ÚnsrrÚ get_cell_len¹s z-DocxTemplate.fix_tables..get_cell_len)r Z XMLParserÚ fromstringÚnsmapÚiterr­ÚfindallÚlenr¯ÚfloatÚsetÚstrr®ÚrangeZ SubElementÚ functoolsÚreducereÚremove)rr,r¨r‡Útr¬ÚcolumnsZto_addrIÚcellsÚwidthZ new_averageÚcZ old_averageÚiZ columns_lenZ cells_len_maxr³Z cells_lenZ to_removeZ removed_widthZ columns_leftZ extra_spacer)r²rr£Žs`      &    4zDocxTemplate.fix_tablescCs@x:|jdtjjjdD]"}|jd7_t|jƒ|jd<qWdS)Nz //wp:docPr)Ú namespacesr Úid)ÚxpathrÚoxmlr²rµr$r»Úattrib)rr‡Úeltrrrr¤äszDocxTemplate.fix_docpr_idscCs| ¡t||ƒS)N)r!r )rZdocpathrrrÚ new_subdocêszDocxTemplate.new_subdocc CsDt|dƒr| ¡}nt|dƒ}| ¡}WdQRXt |¡d@}|S)NÚreadÚrblÿÿ)rdrÍr3ÚbinasciiÚcrc32)Zfile_objÚbufr6ÚcrcrrrÚ get_file_crcîs    zDocxTemplate.get_file_crcc CsL| |¡}t|dƒr$| ¡|j|<n$t|dƒ}| ¡|j|<WdQRXdS)aýReplace one media by another one into a docx This has been done mainly because it is not possible to add images in docx header/footer. With this function, put a dummy picture in your header/footer, then specify it with its replacement in this function using the file path or file-like objects. Syntax: tpl.replace_media('dummy_media_to_replace.png','media_to_paste.jpg') -- or -- tpl.replace_media(io.BytesIO(image_stream), io.BytesIO(new_image_stream)) Note: for images, the aspect ratio will be the same as the replaced image Note2: it is important to have the source media file as it is required to calculate its CRC to find them in the docx rÍrÎN)rÓrdrÍÚcrc_to_new_mediar3)rÚsrc_fileÚdst_filerÒr6rrrÚ replace_mediaùs    zDocxTemplate.replace_mediac CsBt|dƒr| ¡|j|<n$t|dƒ}| ¡|j|<WdQRXdS)aˆReplace embedded picture with original-name given by embedded_file. (give only the file basename, not the full path) The new picture is given by dst_file (either a filename or a file-like object) Notes: 1) embedded_file and dst_file must have the same extension/format in case dst_file is a file-like object, no check is done on format compatibility 2) the aspect ratio will be the same as the replaced image 3) There is no need to keep the original file (this is not the case for replace_embedded and replace_media) rÍrÎN)rdrÍÚpics_to_replacer3)rZ embedded_filerÖr6rrrÚ replace_pics  zDocxTemplate.replace_picc Cs2t|dƒ}| |¡}| ¡|j|<WdQRXdS)aReplace one embedded object by another one into a docx This has been done mainly because it is not possible to add images in docx header/footer. With this function, put a dummy picture in your header/footer, then specify it with its replacement in this function Syntax: tpl.replace_embedded('dummy_doc.docx','doc_to_paste.docx') Note2 : it is important to have the source file as it is required to calculate its CRC to find them in the docx rÎN)r3rÓrÍÚcrc_to_new_embedded)rrÕrÖr6rÒrrrÚreplace_embedded)s  zDocxTemplate.replace_embeddedc Cs(t|dƒ}| ¡|j|<WdQRXdS)a0Replace one file in the docx file First note that a MSWord .docx file is in fact a zip file. This method can be used to replace document embedded in the docx template. Some embedded document may have been modified by MSWord while saving the template : thus replace_embedded() cannot be used as CRC is not the same as the original file. This method works for embedded MSWord file like Excel or PowerPoint file, but won't work for others like PDF, Python or even Text files : For these ones, MSWord generate an oleObjectNNN.bin file which is no use to be replaced as it is encoded. Syntax: tpl.replace_zipname( 'word/embeddings/Feuille_Microsoft_Office_Excel1.xlsx', 'my_excel_file.xlsx') The zipname is the one you can find when you open docx with WinZip, 7zip (Windows) or unzip -l (Linux). The zipname starts with "word/embeddings/". Note that the file is renamed by MSWord, so you have to guess a little bit... rÎN)r3rÍÚzipname_to_replace)rZzipnamerÖr6rrrÚreplace_zipname:s zDocxTemplate.replace_zipnamecCsi|_i|_i|_i|_dS)aZReset replacement dictionaries This will reset data for image/embedded/zipname replacement This is useful when calling several times render() with different image/embedded/zipname replacements without re-instantiating DocxTemplate object. In this case, the right sequence for each rendering will be : - reset_replacements(...) - replace_zipname(...), replace_media(...) and/or replace_embedded(...), - render(...) If you instantiate DocxTemplate object before each render(), this method is useless. N)rÔrÚrÜrØ)rrrrrXszDocxTemplate.reset_replacementsc Csz|js|js|jrvt|dƒr\t ¡}t|ƒ |¡| d¡| d¡|  ¡| d¡nd|}t   ||¡t   |¡Ê}t   |d¡²}xª| ¡D]ž}| |j¡}|j|jkrÆ| ||j|j¡q”|j d¡rô|j|jkrô| ||j|j¡q”|j d¡r&|j|jkr&| ||j|j¡q”| ||¡q”WWdQRXWdQRXt|dƒs`t  |¡t|dƒrv| d¡dS)NrÍrz %s_docxtpl_before_replace_mediasr2z word/media/zword/embeddings/)rÔrÚrÜrdÚioÚBytesIOrÚsaveÚseekÚtruncateÚosÚrenameÚzipfileÚZipFileÚinfolistrÍr5ÚwritestrÚ startswithÚCRCr¿)rZ docx_fileZtmp_fileZzinZzoutÚitemrÑrrrÚpost_processingms:          $   zDocxTemplate.post_processingcCs|jr| ¡dS)N)rØÚ _replace_pics)rrrrÚpre_processing’szDocxTemplate.pre_processingcCsˆdd„|jDƒ}|jj}| ||¡x8t |j¡D](\}}|jtj tj fkr2| |j |¡q2Wx$|  ¡D]\}}|sht d|ƒ‚qhWdS)zRReplaces pictures xml tags in the docx template with pictures provided by the usercSsi|] }d|“qS)Fr)Ú.0Úkeyrrrú šsz.DocxTemplate._replace_pics..z)Picture %s not found in the docx templateN)rØrriÚ_replace_docx_part_picsÚsixÚ iteritemsrœrŒÚREL_TYPEZHEADERZFOOTERrr‹Ú ValueError)rÚ replaced_picsriZrelidr Úimg_idZreplacedrrrrí—s zDocxTemplate._replace_picscCs|jS)N)r")rrrrÚ get_pic_map¬szDocxTemplate.get_pic_mapc Cs®t |j¡}i}|jdtjjjd}xv|D]l}d}yH|jdtjjjdkr˜|jdtjjjdd}|jdtjjjd} t | ƒdkr”| d}qšw,nw,d} |jd | tjjjdd} |jd | tjjjd} | rà| d} nd } |jd | tjjjd}|r |d}nd }|j |j |j |j f|| <xPt  |j¡D]@\}}|| ks`|| ks`||kr:||| d _d||<Pq:WWq,tk r˜w,Yq,Xq,W|j |¡dS)Nz//a:graphic/a:graphicData)rÆrZpiczpic:pic/pic:blipFill/a:bliprz@r:embedzpic:pic/pic:nvPicPr/pic:cNvPr/z%s@namez%s@titler8z%s@descrr T)r r´rŽrÈrrÉr²rµrÊr¸rœZ target_refrrórôrØZ_blobÚ Exceptionr"Úupdate)rZdoc_partr÷ÚetZpart_mapZgdsÚgdr ZblipÚdestZnon_visual_propertiesr5ÚtitlesruZ descriptionsÚ descriptionrøZimg_datarrrrò¯sJ           z$DocxTemplate._replace_docx_part_picscCs| ¡|jjj|tjddS)NT)rŸ)r!rr…Z relate_torõZ HYPERLINK)rÚurlrrrÚ build_url_idæszDocxTemplate.build_url_id)r5rcOsH|js|jst|jƒ|_| ¡|jj|f|ž|Ž| |¡d|_dS)NT)rrr rrrîràrì)rr5ÚargsÚkwargsrrrràës    zDocxTemplate.save)rkrc Csˆ|jdd| ¡}| |¡}xD|j|jgD]4}x.| |¡D] \}}| |¡}|| |¡7}qs.