cCi+SrSSKrSSKJrJr SSKJrJrJr \"5(a SSK r SSK J r \R"\ 5r "SS\5rS/rg) z' Feature extractor class for MarkupLM. N) BatchFeatureFeatureExtractionMixin)is_bs4_availableloggingrequires_backends) BeautifulSoupcL^\rSrSrSrU4SjrSrSrSrS\ 4Sjr S r U=r $) MarkupLMFeatureExtractor!a[ Constructs a MarkupLM feature extractor. This can be used to get a list of nodes and corresponding xpaths from HTML strings. This feature extractor inherits from [`~feature_extraction_utils.PreTrainedFeatureExtractor`] which contains most of the main methods. Users should refer to this superclass for more information regarding those methods. c @>[US/5 [TU]"S0UD6 g)Nbs4)rsuper__init__)selfkwargs __class__s r/home/james-whalen/.local/lib/python3.13/site-packages/transformers/models/markuplm/feature_extraction_markuplm.pyr!MarkupLMFeatureExtractor.__init__+s$( "6"c ^/n/nUR(aUO URmTRHynURTRSS9nUR TR5 UR S[ U5:XaSO[ U4Sj[US5555 UmM{ UR5 UR5 X#4$)NF) recursiverc3:># UHupUTLdM Uv M g7f)Nr).0ischilds r 6MarkupLMFeatureExtractor.xpath_soup..7s 1e@VZ[_dZd!!@Vs  ) nameparentparentsfind_allappendlennext enumeratereverse)relement xpath_tagsxpath_subscriptsr#siblingsrs @r xpath_soup#MarkupLMFeatureExtractor.xpath_soup/s "<<W^^mmFuzzUCH   ejj )  # ##h-'T1e (TU@V1e-e E $   "++rc[US5n/n/n/nURHn[U[RR 5(dM.[ UR5[RRLaM`[R"U5R5nU(dMURU5 URU5upURU5 URU 5 M [U5[U5:wa [S5e[U5[U5:wa [S5eX4U4$)Nz html.parserz3Number of doc strings and xtags does not correspondz3Number of doc strings and xsubs does not correspond)r descendants isinstancerr+NavigableStringtyper#Taghtmlunescapestripr&r/r' ValueError) r html_string html_codeall_doc_stringsstring2xtag_seqstring2xsubs_seqr+text_in_this_tagr,r-s rget_three_from_single.MarkupLMFeatureExtractor.get_three_from_single>s!+}=  ,,G'3;;#>#>??'s{{>#'==#9#?#?#A '&&'78/3w/G, &&z2 ''(89-  3#7 7RS S  3'7#8 8RS S1AAArcdSn[X5HupEUSU3- nUS:wdMUSUS3- nM U$)N/r[])zip)rr,r-xpathtagnamesubss rconstruct_xpath(MarkupLMFeatureExtractor.construct_xpath[sH >MG q ] "Eqy1TF!$? rreturncSn[U[5(aSnOD[U[[45(a)[ U5S:Xd[US[5(aSnU(d[ S[ U5S35e[U[[45=(a [US[5nU(dU/n/n/nUHsnURU5upxn URU5 /n [XxU 5H(upn URX5nU RU5 M* URU 5 Mu XES.n[USS9nU$) ad Main method to prepare for the model one or several HTML strings. Args: html_strings (`str`, `list[str]`): The HTML string or batch of HTML strings from which to extract nodes and corresponding xpaths. Returns: [`BatchFeature`]: A [`BatchFeature`] with the following fields: - **nodes** -- Nodes. - **xpaths** -- Corresponding xpaths. Examples: ```python >>> from transformers import MarkupLMFeatureExtractor >>> page_name_1 = "page1.html" >>> page_name_2 = "page2.html" >>> page_name_3 = "page3.html" >>> with open(page_name_1) as f: ... single_html_string = f.read() >>> feature_extractor = MarkupLMFeatureExtractor() >>> # single example >>> encoding = feature_extractor(single_html_string) >>> print(encoding.keys()) >>> # dict_keys(['nodes', 'xpaths']) >>> # batched example >>> multi_html_strings = [] >>> with open(page_name_2) as f: ... multi_html_strings.append(f.read()) >>> with open(page_name_3) as f: ... multi_html_strings.append(f.read()) >>> encoding = feature_extractor(multi_html_strings) >>> print(encoding.keys()) >>> # dict_keys(['nodes', 'xpaths']) ```FTrzQHTML strings must of type `str`, `list[str]` (batch of examples), but is of type .)nodesxpathsN)data tensor_type) r3strlisttupler'r:r5rAr&rHrLr)r html_strings valid_strings is_batchedrQrRr;r=r>r? xpath_stringsnodetag_listsub_list xpath_stringrSencoded_inputss r__call__!MarkupLMFeatureExtractor.__call__csD`  lC ( ( M  tUm 4 4< A%LOS)I)I $ ""&|"4!5Q8   tUm<c*\Z[_^aBb (>L'KAEA[A[\gAh >O.> LL )M,/Rb,c(#33HG $$\2-d MM- ((1%4TBrr) __name__ __module__ __qualname____firstlineno____doc__rr/rArLrra__static_attributes__ __classcell__)rs@rr r !s1# ,B:T TTrr )rgr7feature_extraction_utilsrrutilsrrrrr get_loggerrcloggerr __all__rrrrosU LAA!   H %V5Vr & &r