a 鴫j@s^ddlmZddlmZddlmZmZddlZddlm Z ddl Z ddl Z ddl m Z ddlZddlZddlmZddlmmZddlZddlmZmZmZmZmZmZmZm Z m!Z!ddl"m#Z$dd l%m&Z&ej'gd d d d Z(ddZ)ddZ*ddZ+ddZ,ej'ej-de.de.dgdej-de.ddgd ddZ/GdddZ0dS))Iterator)partial)BytesIOStringION)Path)URLError)is_platform_windows) NA DataFrame MultiIndexSeries Timestamp date_rangeread_csv read_html to_datetime)file_path_to_url)zchinese_utf-16.htmlzchinese_utf-32.htmlzchinese_utf-8.htmlzletz_latin1.html)paramscCs|ddd|jS)z6Parametrized fixture for HTML encoding test filenames.iodataZ html_encoding)param)requestdatapathrX/var/www/html/laksyashrimp/venv/lib/python3.9/site-packages/pandas/tests/io/test_html.pyhtml_encoding_file$s rcOst|t|ks,Jdt|dt|d}ttdd||}|sPJ|t||D]0\}}tj||g|Ri||jrZJdqZdS)Nz*lists are not of equal size len(list1) == z, len(list2) == z$not all list elements are DataFramescSst|tot|tSN isinstancer )xyrrr:z(assert_framelist_equal..zframes are both empty)lenallmapziptmassert_frame_equalempty)Zlist1Zlist2argskwargsmsgZ both_framesZframe_iZframe_jrrrassert_framelist_equal1s& r-cCsjtd}td||ddtjtdd&t|ddd d dd Wdn1s\0YdS) Nbs4html5lib __version__z4.2zPandas requires versionmatchrrhtml spam.htmlflavor)pytest importorskipsetattrraises ImportErrorr)Z monkeypatchrr.rrrtest_bs4_version_failsEs   r<cCsXd}d}d|d}tjt|d"tt|d|dWdn1sJ0YdS)Nz google.comzinvalid flavorz\{z \} is not a valid set of flavorsr1Zgoogler2r6)r7r: ValueErrorrr)urlr6r,rrrtest_invalid_flavorNs  r@cCsZtdtdtd|dddd}t|ddgd }t|ddgd }t||dS) Nr.lxmlr/rrr3valid_markup.htmlr) index_colr6)r7r8rr-)rfilenameZdfs_lxmlZdfs_bs4rrrtest_same_orderingWs   rEr.r/)ZmarksrAcCstt|jdS)Nr5)rrr)rrrrflavor_read_htmlbsrFc@s8eZdZddZejddZejddZddZd d Z ej j ej j d d Z ej j ej j d dZej jddZddZddZddZddZddZddZddZdd Zd!d"Zd#d$Zd%d&Zd'd(Zd)d*Zd+d,Zd-d.Zd/d0Z d1d2Z!d3d4Z"d5d6Z#d7d8Z$ej j ej j d9d:Z%ej jej j ej j d;d<Z&ej jd=d>Z'ej jd?d@Z(ej jdAdBZ)ej jdCdDZ*ej jdEdFZ+ej jdGdHZ,ej jdIdJZ-ej jdKdLZ.ej jdMdNZ/dOdPZ0ejdQdRZ1ej j ej j dSdTZ2ej j ej j dUdVZ3dWdXZ4dYdZZ5d[d\Z6d]d^Z7d_d`Z8dadbZ9ej jdcddZ:ej jdedfZ;dgdhZdmdnZ?dodpZ@dqdrZAdsdtZBdudvZCdwdxZDdydzZEd{d|ZFd}d~ZGddZHej IdddgddZJddZKddZLddZMddZNddZOddZPddZQej jddZRddZSddZTej IddeUdgdfdeUdgeUdgfgddZVej IdddgddZWeXYej ZdddZ[ddZ\ddZ]ej jej j ddZ^ddZ_ddZ`ej IdgdddZaddZbddZcddZdddZedS) TestReadHtmlcCs>d}tjt|d|dWdn1s00YdS)NzPassing literal html to 'read_html' is deprecated and will be removed in a future version. To read from a literal string, wrap it in a 'StringIO' object.r1a
A B
1 2
3 4
)r'Zassert_produces_warning FutureWarning)selfrFr,rrrtest_literal_html_deprecationms z*TestReadHtml.test_literal_html_deprecationcCs|ddddS)Nrrr3r4rrIrrrr spam_dataszTestReadHtml.spam_datacCs|ddddS)Nrrr3 banklist.htmlrrKrrr banklist_dataszTestReadHtml.banklist_datacCsdttjddttdddj t }| }|t |ddidd d}t ||dS) N)abccolumnsz{:.3f}classZ dataframer)attrsrC)r nprandom default_rngpdIndexlistr%formatZastypefloatto_htmlrr'r()rIrFdfoutresrrrtest_to_html_compats  z TestReadHtml.test_to_html_compatc sttdtjdgddtgdddtdtjdgddtgd ddgd gd gd gd d}|jdd}td|$|t||dd}Wdn1s0Y|dkrt dt  }n t |}ttdtjdgddtgdddtdtjdgddtgd ddtddt gddtgd ddtgd |dtgd |dd|dkrddlddlmtfddjDtj|dddS)NrQZInt64)dtype)rdrOrQ?@ZFloat64)rfg@rg)TFN)TFTabc)rirjN)rirjrkdefghFindexzmode.string_storage dtype_backendrpyarrowTboolean)ArrowExtensionArraycs$i|]}|j|ddqS)T)Z from_pandas)array).0colrwexpectedparr sz3TestReadHtml.test_dtype_backend..)Zcheck_column_type)r r rWnanr_rZZoption_contextrr7r8Z ArrowDtypestringZ StringDtyper ruZ pandas.arraysrwrTr'r()rIZstring_storagertrFr`raresultZ string_dtyperr{rtest_dtype_backendsL 2     zTestReadHtml.test_dtype_backendcCsft|dd<}|j|d||jdd}||jdd}Wdn1sN0Yt||dS)Nutf-8encodingcontentFirst Federal Bank of Floridar1 Metcalf Bankopen serve_contentreadr?r-)rI httpserverrNrFrndf1df2rrrtest_banklist_urls$zTestReadHtml.test_banklist_urlcCsft|dd<}|j|d||jdd}||jdd}Wdn1sN0Yt||dS)Nrrr .*Water.*r1Unitr)rIrrLrFrnrrrrr test_spam_urls ,zTestReadHtml.test_spam_urlcCs2||dddid}||dddid}t||dS)Nz .*Florida.*idtabler2rVrr-)rIrNrFrrrrr test_banklists  zTestReadHtml.test_banklistcCsR||dd}||dd}t|||djddks8J|djddksNJdS)Nrr1rrrr ProximatesZNutrient)r-ilocrTrIrLrFrrrrr test_spams    zTestReadHtml.test_spamcCs$||}|D]}t|ts Jq dSrr)rIrLrFdfsr`rrrtest_spam_no_match szTestReadHtml.test_spam_no_matchcCs,||ddid}|D]}t|tsJqdS)Nrr)rVr)rIrNrFrr`rrrtest_banklist_no_matchsz#TestReadHtml.test_banklist_no_matchcCs2||dddd}|jddks$J|jr.JdS)NrrOr2headerrr)rTr))rIrLrFr`rrrtest_spam_headerszTestReadHtml.test_spam_headercCs*||ddd}||ddd}t||dSNrrdr2skiprowsrrrrrrtest_skiprows_intszTestReadHtml.test_skiprows_intcCs2||dtdd}||dtdd}t||dSNrrOrr)ranger-rrrrtest_skiprows_rangesz TestReadHtml.test_skiprows_rangecCs2||dddgd}||dddgd}t||dSNrrdrOrrrrrrrtest_skiprows_list$szTestReadHtml.test_skiprows_listcCs2||dddhd}||dddhd}t||dSrrrrrrtest_skiprows_set*szTestReadHtml.test_skiprows_setcCs*||ddd}||ddd}t||dSrrrrrrtest_skiprows_slice0sz TestReadHtml.test_skiprows_slicecCs2||dtdd}||dtdd}t||dSrslicer-rrrrtest_skiprows_slice_short6sz&TestReadHtml.test_skiprows_slice_shortcCs8||dtddd}||dtdddd}t||dS) NrrOrrrPrdrrrrrtest_skiprows_slice_long<sz%TestReadHtml.test_skiprows_slice_longcCs6||dtdd}||dtdd}t||dSr)rWZaranger-rrrrtest_skiprows_ndarrayBsz"TestReadHtml.test_skiprows_ndarraycCs@tjtdd||dddWdn1s20YdS)Nz%is not a valid type for skipping rowsr1rZasdfr)r7r: TypeError)rIrLrFrrrtest_skiprows_invalidHsz"TestReadHtml.test_skiprows_invalidcCs*||ddd}||ddd}t||dSNrrr2rCrrrrrr test_indexLszTestReadHtml.test_indexcCs.||dddd}||dddd}t||dSNrrdr)r2rrCrrrrrrtest_header_and_index_no_typesQsz+TestReadHtml.test_header_and_index_no_typescCs.||dddd}||dddd}t||dSrrrrrr test_header_and_index_with_typesVsz-TestReadHtml.test_header_and_index_with_typescCs*||ddd}||ddd}t||dSrrrrrrtest_infer_types[szTestReadHtml.test_infer_typescCst|dd}t|}Wdn1s.0Yt|dd}t|}Wdn1sf0Y||dd}||dd}t||dSNzUTF-8rrr1r)rrrr-)rIrLrFrndata1data2rrrrrtest_string_ioas**  zTestReadHtml.test_string_iocCsbt|dd}|}Wdn1s*0Y|t|dd}|t|dd}t||dSr)rrrr-)rIrLrFrnrrrrrr test_stringls &zTestReadHtml.test_stringcCs~t|dd}||dd}Wdn1s.0Yt|dd}||dd}Wdn1sf0Yt||dSr)rr-)rIrLrFrnrrrrrtest_file_likeus **zTestReadHtml.test_file_likecCsL|jdddtjtdd|dddWdn1s>0YdS)Nz#urlopen error unknown url type: gitcoder1zgit://github.comr)rr7r:rrIrrFrrrtest_bad_url_protocol~sz"TestReadHtml.test_bad_url_protocolcCsR|jdddtjttfdd||jddWdn1sD0YdS)NzName or service not knownrrzHTTP Error 404: NOT FOUNDr1r)rr7r:rr>r?rrrrtest_invalid_urlszTestReadHtml.test_invalid_urlcCsL|}|ttj|dddid}t|ts0J|D]}t|ts4Jq4dS)NZFirstrrr)rospathabspathrr\r rIrNrFr?rr`rrr test_file_urlszTestReadHtml.test_file_urlcCsH|}tjtdd"||dddidWdn1s:0YdS)NzNo tables foundr1rrZ tasdfablerr7r:r>)rIrNrFr?rrrtest_invalid_table_attrss  z%TestReadHtml.test_invalid_table_attrscCs0||dddiddgdd}t|jts,JdS)NMetcalfrrrrd)r2rVrrrTr rIrNrFr`rrrtest_multiindex_headers z#TestReadHtml.test_multiindex_headercCs0||dddiddgdd}t|jts,JdS)Nrrrrrd)r2rVrC)rrrr rrrrtest_multiindex_indexs z"TestReadHtml.test_multiindex_indexcCsF||dddiddgddgdd}t|jts2Jt|jtsBJdS)Nrrrrrd)r2rVrrC)rrTr rrrrrrtest_multiindex_header_indexsz)TestReadHtml.test_multiindex_header_indexcCs2||dddiddgddd}t|jts.JdSNrrrrrd)r2rVrrrrrrr&test_multiindex_header_skiprows_tuplessz3TestReadHtml.test_multiindex_header_skiprows_tuplescCs2||dddiddgddd}t|jts.JdSrrrrrrtest_multiindex_header_skiprowssz,TestReadHtml.test_multiindex_header_skiprowscCsH||dddiddgddgddd}t|jts4Jt|jtsDJdS)Nrrrrrd)r2rVrrCr)rrrr rTrrrr%test_multiindex_header_index_skiprowssz2TestReadHtml.test_multiindex_header_index_skiprowscCsX|}|ttj|ttdddid}t|ts

What's new in Python 2.7?
or all "What's new" documents since 2.0

Tutorial
start here

Library Reference
keep this under your pillow

Language Reference
describes syntax and language elements

Python Setup and Usage
how to use Python on different platforms

Python HOWTOs
in-depth documents on specific topics

Installing Python Modules
installing from the Python Package Index & other sources

Distributing Python Modules
publishing modules for installation by others

Extending and Embedding
tutorial for C/C++ programmers

Python/C API
reference for C/C++ programmers

FAQs
frequently asked questions (with answers!)

Indices and tables:

Python Global Module Index
quick access to all modules

General Index
all functions, classes, terms

Glossary
the most important terms explained

Search page
search this documentation

Complete Table of Contents
lists all sections and subsections

rrIrrr python_docsszTestReadHtml.python_docscCs.|j|d||jdd}t|dks*JdS)NrPythonr1rd)rr?r#)rIrrrFrrrrtest_multiple_matchess z"TestReadHtml.test_multiple_matchescCs@|j|d||jdd}dd|D}t|ddgks*r"z7TestReadHtml.test_python_docs_table..ZPythZWhat)rr?sorted)rIrrrFrzzrrrtest_python_docs_table%s z#TestReadHtml.test_python_docs_tablecCs$d}|t|}t|dks JdS)z@ Make sure that read_html ignores empty tables. a
A B
1 2
rdN)rr#)rIrFr3rrrrtest_empty_tables-s zTestReadHtml.test_empty_tablescCs<|tdd}tddgddggddgd }t||dS) Na
A B
1 2
3 4
rrdrOrQrPABrrTrr r'r(rIrFrr|rrrtest_multiple_tbodyHsz TestReadHtml.test_multiple_tbodycCs2|tdd}tddidgd}t||dS)zt Don't fail with bs4 when there is a header and only one column as described in issue #9178 a3
Header
first
rHeaderfirstrrrNrrrrrtest_header_and_one_columnhsz'TestReadHtml.test_header_and_one_columncCs6|tdd}tgdggdd}t||dS)zK Ensure parser adds within on malformed HTML. a
Country Municipality Year
Ukraine Odessa 1944
r)ZUkraineZOdessa)ZCountryZ MunicipalityZYearrNrrrrrtest_thead_without_trsz"TestReadHtml.test_thead_without_trc Csd}tddggddgd}tddgddggddgd}|jd d }|jd d }|t|d }|t|d }t||t||d S)zh Make sure that read_html reads tfoot, containing td or th. Ignores empty tfoot a {footer}
A B
bodyA bodyB
ZbodyAZbodyBrrrZfootAZfootB)footerz%footAfootBrN)r r]rr'r() rIrFZ data_templateZ expected1Z expected2rrZresult1Zresult2rrrtest_tfoot_reads   zTestReadHtml.test_tfoot_readcCs6|tdddd}tddggdd}t||dS)Na 
S I
text 1944
rrtextr)SIrSrrrrr&test_parse_header_of_non_string_columnsz3TestReadHtml.test_parse_header_of_non_string_columnc sddlmfdd}||dddidd}t|d d d d ttd d}|j|jksZJgd}gd}||||} ||} | } ddg} | | t| | <t | | dS)Nr_remove_whitespacecs&z |WSty |YS0dSr)AttributeErrorrrrr try_remove_wss  z8TestReadHtml.test_banklist_header..try_remove_wsrrrrrrcsvz banklist.csv) Updated Date Closing Date converters) z,First Vietnamese American Bank In Vietnamesez"Westernbank Puerto Rico En Espanolz*R-G Premier Bank of Puerto Rico En EspanolzEurobank En EspanolzSanderson State Bank En EspanolzLWashington Mutual Bank (Including its subsidiary Washington Mutual Bank FSB)zSilver State Bank En Espanolz%AmTrade International Bank En EspanolzHamilton Bank, NA En Espanolz6The Citizens Savings Bank Pioneer Community Bank, Inc.) zFirst Vietnamese American BankzWesternbank Puerto RicozR-G Premier Bank of Puerto RicoZEurobankzSanderson State BankzWashington Mutual BankzSilver State BankzAmTrade International BankzHamilton Bank, NAzThe Citizens Savings Bankrr) Zpandas.io.htmlrrr shaper%replaceapplyrr'r() rIrNrrFrr`Z ground_trutholdnewZdfnewZgtnewZ convertedZ date_colsrrrtest_banklist_headers       z!TestReadHtml.test_banklist_headercCsnd}t|dd}|}Wdn1s.0Y||vsDJ||dddidd}||vsjJdS)Nz Gold Canyonrrrrrr)rrZ to_string)rIrNrFgcrnZraw_textr`rrrtest_gold_canyons&  zTestReadHtml.test_gold_canyoncCs8|tdddd}|tdddd}t||dS)Na
C_l0_g0 C_l0_g1 C_l0_g2 C_l0_g3 C_l0_g4
R_l0_g0 0.763 0.233 nan nan nan
R_l0_g1 0.244 0.285 0.392 0.137 0.222
rrCa
C_l0_g0 C_l0_g1 C_l0_g2 C_l0_g3 C_l0_g4
R_l0_g0 0.763 0.233
R_l0_g1 0.244 0.285 0.392 0.137 0.222
)rr'r()rIrFr|rrrrtest_different_number_of_colss" "$!z*TestReadHtml.test_different_number_of_colscCs6|tdd}tgdggdd}t||dS)NaZ
A B C
a b c
rrh)rrCrSrrrrrtest_colspan_rowspan_1esz#TestReadHtml.test_colspan_rowspan_1cCs:|tdddd}tgdggdd}t||dS)Na
X Y Z W
A B C
rr)rrrZr)XzX.1YrWrrrrrr test_colspan_rowspan_copy_values~sz-TestReadHtml.test_colspan_rowspan_copy_valuescCs:|tdddd}tgdggdd}t||dS)Na(
A B C
D
rr)rrrrD)rrzB.1zB.2rrrrrrrtest_colspan_rowspan_both_not_1sz,TestReadHtml.test_colspan_rowspan_both_not_1cCs:|tdddd}tddggddgd}t||dS)Nz
A B
C
rrrrrrrrrrrtest_rowspan_at_end_of_rows z'TestReadHtml.test_rowspan_at_end_of_rowcCs@|tdddd}tddgddggddgd}t||dS)Nz
A B
rrrrrrrrrrtest_rowspan_only_rowss  z#TestReadHtml.test_rowspan_only_rowscCsV|tdd}tddgddggddgddggd}tdd gg|d }t||dS) Nam
A B
a b
1 2
rrrrirjrdZlevelscodesrOrrr r r'r(rIrFrrTr|rrr+test_header_inferred_from_rows_with_only_ths$z8TestReadHtml.test_header_inferred_from_rows_with_only_thcCshtdtdddi}|}|t|dgdd}t||d|t|dgdd}t||ddS)Ndate1/1/2001 ZperiodsrdrZ parse_datesrC)r rr_rr'r()rIrFr`r|rbrrrtest_parse_dates_lists z"TestReadHtml.test_parse_dates_listcCspttddd}t|dd|ddd}|t|dd d gid d }td|i}t||d dS) Nr#r$r%cSs t|Sr)strr"rrrrr!r"z7TestReadHtml.test_parse_dates_combine..cSs t|Sr)r(timerrrrr!r")r"r)datetimerdrOr&r)r rr r%rr_r'r()rIrFZ raw_datesr`rbZnewdfrrrtest_parse_dates_combines   z%TestReadHtml.test_parse_dates_combinecCs|dddd}tj|s,Jt|dtj|sJJt|d||ddd d }|jd ksjJd |jd vs|J|djtdksJt |j ddsJdS)Nrrr3wikipedia_states.htmlz is not a filez is an empty fileArizonardrr)< Unnamedrsq mifloat64)rr1HzPN$A) rrisfilereprgetsizerrTrerWallcloselocrIrrFrrrrrtest_wikipedia_states_table%sz(TestReadHtml.test_wikipedia_states_tablecCsn|dddd}||dddd}|jdks.Jd |jd d vsDJ|jjd ksTJt|jd dsjJdS)Nrrr3r,r-rr)r. r0rrdrO)ZAlaska)z Total area[2]r1r3)rrTZnlevelsrWr7r8r9rrr test_wikipedia_states_multiindex/s z-TestReadHtml.test_wikipedia_states_multiindexcCsD|tdddgd}tddggtddgd }t|d|dS) NaK
AB
ab
rrdrrirj)Unnamed: 0_level_0r)zUnnamed: 1_level_0rrS)rr r from_tuplesr'r(rrrr%test_parser_error_on_empty_header_row7s z2TestReadHtml.test_parser_error_on_empty_header_rowcCsN|tdddd}tddidgd}|djtdks>Jt||dS) Na
Header
1100#101
#)decimalrrgClg0@rr2)rr rerWr'r(rrrrtest_decimal_rowsPszTestReadHtml.test_decimal_rowsargTFcCsHtd}tjt|d|||dWdn1s:0YdS)NzPassing a bool to header is invalid. Use header=None for no header or header=int or list-like of ints to specify the row(s) making up the column namesr1r)rescaper7r:r)rIrLrCrFr,rrrtest_bool_header_argms z!TestReadHtml.test_bool_header_argcCs8|tddtidd}tdddgi}t||dS)Na
a
0.763
0.244
rirrz0.763z0.244)rr(r r'r(rrrrtest_convertersxszTestReadHtml.test_converterscCs8|tddgdd}tddtjgi}t||dS)Na
a
0.763
0.244
gZd;?)Z na_valuesrrig"~j?rr rWrr'r(rrrrtest_na_valuesszTestReadHtml.test_na_valuescCsld}tdddgi}|t|ddd}t||tdtjtjgi}|t|ddd}t||dS) Na
a
N/A
NA
rizN/Ar F)Zkeep_default_narT)r rr'r(rWr)rIrFZ html_data expected_dfhtml_dfrrrtest_keep_default_nas z!TestReadHtml.test_keep_default_nacCs@|tdd}tddgtjtjggddgd}t||dS)Nak
A B
a b
rrirjrrrrGrrrrtest_preserve_empty_rowss z%TestReadHtml.test_preserve_empty_rowscCsV|tdd}tddgddggddgddggd}tdd gg|d }t||dS) NaU
AB
ab
12
rrrrirjrdrrOrrr rrr,test_ignore_empty_rows_when_inferring_headers$z9TestReadHtml.test_ignore_empty_rows_when_inferring_headercCsLtgdd}gdgdg|_|jdd}|t|d}t||dS)N))ZHillaryDr)ZBernieJr)ZDonaldER)r)r=ZAgeZParty)NamezUnnamed: 1_level_1zUnnamed: 2_level_1Frqr)r rTr_rr'r()rIrFrIr3rJrrrtest_multiple_header_rowss z&TestReadHtml.test_multiple_header_rowscCs>|dddd}||dd}t|ts(Jt|dts:JdS)Nrrr3rBrr)rr\r )rIrrFrDrrrrtest_works_on_valid_markups z'TestReadHtml.test_works_on_valid_markupcCs$|dddd}||dddgddS) Nrrr3rMrrAr/r=r)rIrrFrNrrrtest_fallback_success sz"TestReadHtml.test_fallback_successcCs>tddd}ttjdd|d}|}d|vs:JdS)Nz 2000-01-01r$r%rO)r$rPrq)rr rWrXrYZstandard_normalr_)rIrngr`rrrrtest_to_html_timestamps z#TestReadHtml.test_to_html_timestampcCstdddg}|}|jdd}|jdd}|jdd}|jdd}|jdd}d|vs`J||kslJ||ksxJ||ksJd |vsJd |vsJd |vsJ||ksJdS) NrdrOrrT)borderrFz border="1"z border="2"z border="0"z border)r r_)rIr`Zout_border_defaultZout_border_trueZout_border_explicit_defaultZout_border_nondefaultZout_border_zeroZout_border_falserrrtest_to_html_borderlesss            z$TestReadHtml.test_to_html_borderlesszdisplayed_only,exp0,exp1ZfooNzfoo bar baz quxcCsRd}|t||d}t|d||dur>t|d|nt|dksNJdS)Na
foo bar baz qux
foo
displayed_onlyrrd)rr'r(r#)rIr\Zexp0Zexp1rFrrrrrtest_displayed_only+s z TestReadHtml.test_displayed_onlyr\cCs>d}|t||dd}tddgddgd}t||dS) NaW
A B
1 2
4 5
r[rrdrPrOrrXr)rIr\rFZ html_tablerr|rrr&test_displayed_only_with_many_elementsPs z3TestReadHtml.test_displayed_only_with_many_elementsz\ignore:You provided Unicode markup but also provided a value for from_encoding.*:UserWarningc Cstj|}tj|d}|d\}}zt|d&}|||dd}Wdn1sb0Yt|d*}|t||dd} Wdn1s0Y|||dd} t || t || Wn:t yt rd|vsd|vrt Yn0dS)Nr_rb)rrCZ16Z32)rrbasenamesplitextsplitrrpoprr'r( Exceptionrr7skip) rIrrF base_pathrootr_rZfobjZ from_stringZfrom_file_like from_filenamerrr test_encodehs,   ( (  zTestReadHtml.test_encodecCsx|jddkrtdGdddt}|d}||s>Jtjtdd||Wdn1sj0YdS) Nr6rAzNot applicable for lxmlc@seZdZddZdS)zFTestReadHtml.test_parse_failure_unseekable..UnseekableStringIOcSsdSNFrrrrrseekableszOTestReadHtml.test_parse_failure_unseekable..UnseekableStringIO.seekableN)__name__ __module__ __qualname__rlrrrrUnseekableStringIOsrpz?
spameggs
z#passed a non-rewindable file objectr1)keywordsgetr7rfrr:r>)rIrFrpbadrrrtest_parse_failure_unseekables  z*TestReadHtml.test_parse_failure_unseekablecCs:Gddd}|d}|d}||s*J||s6JdS)Nc@sJeZdZddddZdddZddZd d Zd d Zedd dZ dS)z9TestReadHtml.test_parse_failure_rewinds..MockFileN)returncSs||_d|_dSrk)rat_end)rIrrrr__init__szBTestReadHtml.test_parse_failure_rewinds..MockFile.__init__cSs|jr dn|j}d|_|S)NrT)rvr)rIsizerrrrrsz>TestReadHtml.test_parse_failure_rewinds..MockFile.readcSs d|_dSrk)rv)rIoffsetrrrseeksz>TestReadHtml.test_parse_failure_rewinds..MockFile.seekcSsdS)NTrrrrrrlszBTestReadHtml.test_parse_failure_rewinds..MockFile.seekablecSsdSrrrrrr__next__szBTestReadHtml.test_parse_failure_rewinds..MockFile.__next__cSs|Srrrrrr__iter__szBTestReadHtml.test_parse_failure_rewinds..MockFile.__iter__)N) rmrnrorwrrzrlr{rr|rrrrMockFiles  r}z/
spam
eggs
z2
spameggs
r)rIrFr}Zgoodrsrrrtest_parse_failure_rewindss  z'TestReadHtml.test_parse_failure_rewindscCsGdddtj}|dddd}|||fd}|||fd}|||sL|r^qLd|jurv|jus|nJdS)NcseZdZfddZZS)z@TestReadHtml.test_importcheck_thread_safety..ErrorThreadc sDztWn*ty8}z||_WYd}~nd}~00d|_dSr)superrunreerr)rIr __class__rrrs zDTestReadHtml.test_importcheck_thread_safety..ErrorThread.run)rmrnror __classcell__rrrr ErrorThreadsrrrr3rB)targetr*) threadingThreadstartis_aliver)rIrrFrrDZhelper_thread1Zhelper_thread2rrrtest_importcheck_thread_safetys z+TestReadHtml.test_importcheck_thread_safetycCs>|dddd}t|}||d}||d}t||dS)Nrrr3r4r)rr'r()rIrrFZfile_path_string file_pathrrrrrtest_parse_path_objects   z#TestReadHtml.test_parse_path_objectcCs2|tdd}tdggdgd}t||dS)Nz
A
word1
word2
rz word1 word2rrrrrrrtest_parse_br_as_spaces z#TestReadHtml.test_parse_br_as_space)r$bodyrrc Csd}gdgdgdgdgdgdd}|d }|d }|d }|d krh|d }|d}|d}n4|dkrz|d }n"|dkr|d}n|dkr|d}|t||dd}t||g|d} | tj} t|| dS)Na
HTTP FTP Linkless
Wikipedia SURROUNDING Debian TEXT Linkless
Footer Multiple links: Only first captured.
)HTTPFTPLinkless))rN)rN)rz'https://en.wiktionary.org/wiki/linkless) WikipediaSURROUNDING Debian TEXTr))rzhttps://en.wikipedia.org/)rzftp://ftp.us.debian.org/)rN)Footer$Multiple links: Only first captured.N))rz)https://en.wikipedia.org/wiki/Page_footer)r1N) head_ignore head_extract body_ignore body_extract footer_ignorefooter_extractrrrr$rrrrrrZ extract_linksrrS)rr ZfillnarWrr'r() rIrCrFZ gh_13141_dataZgh_13141_expectedZdata_expZfoot_expZhead_exprr|rrrtest_extract_linkss2    zTestReadHtml.test_extract_linkscCsBd}tjt|dt|ddWdn1s40YdS)NzY`extract_links` must be one of {None, "header", "footer", "body", "all"}, got "incorrect"r1Z incorrectrr7r:r>r)rIrLr,rrrtest_extract_links_bad:sz#TestReadHtml.test_extract_links_badcCs4d}|t|ddd}tdgg}t||dS)Nz
Google.com
r$rr)z Google.comzhttps://google.comrrIrFrrr|rrr test_extract_links_all_no_headerBs  z-TestReadHtml.test_extract_links_all_no_headercCsBd}tjt|dtdddWdn1s40YdS)NzPdtype_backend numpy is invalid, only 'numpy_nullable' and 'pyarrow' are allowed.r1testnumpyrsr)rIr,rrrtest_invalid_dtype_backendQsz'TestReadHtml.test_invalid_dtype_backendcCs@d}|t|d}tddgddggddgd }t||dS) Na
A B
A1 B1
A2 B2
rA1ZB1A2ZB2rrrrrrrrtest_style_tagYszTestReadHtml.test_style_tag)frmrnrorJr7fixturerLrNrcrmarknetworkZ single_cpurrZslowrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr rrrrrrrr!r'r+r:r<r?rBZ parametrizerErFrHrKrLrMrSrTrUrWrZr r]r^tdZskip_if_windowsfilterwarningsrjrtr~rrrrrrrrrrrrrGls   6                 0    & / H#        "  DrG)1collections.abcr functoolsrrrrrpathlibrrr urllib.errorrrrWr7Z pandas.compatrZpandas.util._test_decoratorsutilZ_test_decoratorsrZpandasrZr r r r r rrrrZpandas._testingZ_testingr'Zpandas.io.commonrrrr-r<r@rErZ skip_if_norFrGrrrrs<     ,