o Zh0hF-@sdZddlZddlmZddlmZmZmZddlmZddl m Z ddl m Z dd l mZdd lmZdd lmZe Zeed d Zeedd Zeedd ZeeedZeeedZeedZdZededZddZddZiZ ddZ!ddZ"ddZ#e"dd d!Z$e"d"d#d$Z%e"d%d&d'Z&e"d(d)d*Z'e"d+d,d-Z(e"d.d/d0Z)d1d2Z*d3d4Z+ed5ej,Z-d6d7Z.dS)8aVSpecific handling for some HTML elements, especially replaced elements. Replaced elements (eg. elements) are rendered externally and behave as an atomic opaque box in CSS. In general, they may or may not have intrinsic dimensions. But the only replaced elements currently supported in WeasyPrint are images with intrinsic dimensions. N)files)CSS Attachmentcss)get_child_text) CounterStyle)boxes)SVGImage)LOGGER)get_url_attributez html5_ua.csszutf-8zhtml5_ua_form.cssz html5_ph.css)string counter_styler z z[^z]+cCs|S)a4Transform (only) ASCII letters to lower case: A-Z is mapped to a-z. This is used for `ASCII case-insensitive `_ matching. This is different from the :meth:`str.lower` method of Unicode strings which also affect non-ASCII characters, sometimes mapping them into the ASCII range: >>> keyword = 'Bac\N{KELVIN SIGN}ground' >>> assert keyword.lower() == 'background' >>> assert ascii_lower(keyword) != keyword.lower() >>> assert ascii_lower(keyword) == 'bac\N{KELVIN SIGN}ground' )encodelowerdecoderrE/var/www/html/rh/venv/lib/python3.10/site-packages/weasyprint/html.py ascii_lower$srcs(t|dd}tfdd|DS)zDReturn whether element has a ``rel`` attribute with given link type.relc3s|] }t|kVqdS)N)r).0token link_typerr ;sz(element_has_link_type..)HTML_SPACE_SEPARATED_TOKENS_REfindallgetany)elementrtokensrrrelement_has_link_type8sr#cCs$|jtvrt|j||||S|gS)zbHandle HTML elements that need special care. :returns: a (possibly empty) list of boxes. ) element_tag HTML_HANDLERStag)r!boxget_image_from_uribase_urlrrrhandle_elementBs r*csfdd}|S)zDReturn a decorator registering a function handling ``tag`` elements.cs |t<|S)z;Decorator registering a function handling ``tag`` elements.)r%)functionr&rr decoratorPszhandler..decoratorr)r&r-rr,rhandlerNs r.cCs@d|jdvr tjntj}||j|j||}|j|_|j|_|S)zWrap an image in a replaced box. That box is either block-level or inline-level, depending on what the element should be. blockdisplay)styler BlockReplacedBoxInlineReplacedBoxr& string_setbookmark_label)r!r'imagetype_new_boxrrrmake_replaced_boxWsr9imgcCst|d|}|d}|r=|||jdd}|dur!t|||gS|r/tj||g|_|gS|dkr5gS|dus;JgS|rKtj||g|_|gSgS)zHandle ```` elements. Return either an image or the alt-text. See: https://www.w3.org/TR/html5/embedded-content-1.html#the-img-element srcaltimage_orientation)url orientationNr)r rr1r9r TextBoxanonymous_fromchildren)r!r'r(r)r;r<r6rrr handle_imgis&   rCembedcCsNt|d|}|dd}|r%||||jdd}|dur%t|||gSgS)zHandle ```` elements, return either an image or nothing. See: https://www.w3.org/TR/html5/embedded-content-0.html#the-embed-element r;typerr=r>forced_mime_typer?Nr rstripr1r9)r!r'r(r)r;r7r6rrr handle_embeds rJobjectcCsPt|d|}|dd}|r%||||jdd}|dur%t|||gS|gS)zHandle ```` elements, return either an image or the fallback. See: https://www.w3.org/TR/html5/embedded-content-0.html#the-object-element datarErr=rFNrH)r!r'r(r)rLr7r6rrr handle_objects rMcolgroupcs>ttjrtdd|DsfddtjD_gS)Handle the ``span`` attribute.css|]}|jdkVqdS)colNr,)rchildrrrrsz"handle_colgroup..csg|] }tjgqSr)r TableColumnBoxrA)r_r'rr s z#handle_colgroup..) isinstancer TableColumnGroupBoxr rangespanrBr!r'_get_image_from_uri _base_urlrrTrhandle_colgroups  r]rPcs4ttjrjdkrfddtjDSgS)rOrcsg|]}qSr)copy)r_irTrrrUszhandle_col..)rVr rRrYrXrZrrTr handle_colsr`z{http://www.w3.org/2000/svg}svgc Csx|jd}|jd}z t||||}Wn!ty4}ztd|tjd|dgWYd}~Sd}~wwt|||gS)zUHandle ```` elements. Return either an image or the fallback content. url_fetchercontextzFailed to load inline SVG: %szError while loading inline SVG:)exc_infoN)keywordsr ��� Exceptionr ���errordebugr9���)r!���r'���r(���r)���ra���rb���r6��� exceptionr���r���r��� handle_svg���s���   ri���c�������������� ���C���s��d}d}d}g�}g�}d}d}g�}i�} |�j�jdd} |�jdddD�]} | j�} | jdkr5|du�r5t| }q"| jdkrt| dd} | dd} | d kratt | d D�] }||vr_| |�qTq"| d krk| | �q"| d krv|du�ru| }q"| d kr|du�r| }q"| dkr|du�rt | | }q"| dkr|du�rt | | }q"| r| | vr| | | <�q"| jdkrt | drt| d|�j}| dd}|du�rtd�q"t|||�jd}| |�q"||||||||| | d S�)a2��Get metadata dictionary out of HTML object. Relevant specs: https://www.whatwg.org/html#the-title-element https://www.whatwg.org/html#standard-metadata-names https://wiki.whatwg.org/wiki/MetaExtensions https://microformats.org/wiki/existing-rel-values#HTML5_link_type_extensions Nlangtitlemetalinknamer���contentrd���,author description generatorzdcterms.createdzdcterms.modified attachmenthrefz'Missing href in )r>���rr���ra���) rk���rr���rs���rd���authorscreatedmodified attachmentsrj���custom) etree_elementattribr���wrapper_element query_allr&���r���r���mapstrip_whitespacesplitappendparse_w3c_dater#���r ���r)���r ���rf���r���ra���)htmlrk���rr���rs���rd���rv���rw���rx���ry���rz���rj���r!���rn���ro���keywordr>���attachment_titlert���r���r���r���get_html_metadata���s���             r���c�����������������C���s ���|��tS�)zUse the HTML definition of "space character", not all Unicode Whitespace. https://www.whatwg.org/html#strip-leading-and-trailing-whitespace https://www.whatwg.org/html#space-character )rI���HTML_WHITESPACEr���r���r���r���r���&��s��� r���aG�� ^ [ ]* (?P\d\d\d\d) (?: -(?P0\d|1[012]) (?: -(?P[012]\d|3[01]) (?: T(?P[01]\d|2[0-3]) :(?P[0-5]\d) (?: :(?P[0-5]\d) (?:\.\d+)? # Second fraction, ignored )? (?: Z | # UTC (?P[+-](?:[01]\d|2[0-3])) :(?P[0-5]\d) ) )? )? )? [ ]* $ c�����������������C���s ���t�|r|S�td|�|�dS�)zYParse datetimes as defined by the W3C. See https://www.w3.org/TR/NOTE-datetime z#Invalid date in %rN) W3C_DATE_REmatchr warning) meta_namer rrrrTs r)/__doc__reimportlib.resourcesrrrrrr css.countersrformatting_structurer imagesr loggerr urlsr HTML5_UA_COUNTER_STYLE read_textHTML5_UA HTML5_UA_FORMHTML5_PHHTML5_UA_STYLESHEETHTML5_UA_FORM_STYLESHEETHTML5_PH_STYLESHEETrcompilerrr#r%r*r.r9rCrJrMr]r`rirrVERBOSErrrrrrs^           $     I