o թZh@snddlmZddlmZddlmZmZddlZddlm Z ddl Z ddl Z ddl m Z ddlZddlZddlmZddlmmZddlZddlmZmZmZmZmZmZmZm Z m!Z!ddl"m#Z$dd l%m&Z&m'Z'dd l(m)Z)ej*gd d d dZ+ddZ,ddZ-ddZ.ddZ/ej*ej0de1de1dgdej0de1ddgd ddZ2GdddZ3dS))Iterator)partial)BytesIOStringION)Path)URLError)is_platform_windows) NA DataFrame MultiIndexSeries Timestamp date_rangeread_csv read_html to_datetime)ArrowStringArray StringArray)file_path_to_url)zchinese_utf-16.htmlzchinese_utf-32.htmlzchinese_utf-8.htmlzletz_latin1.html)paramscCs|ddd|jS)z6Parametrized fixture for HTML encoding test filenames.iodataZ html_encoding)param)requestdatapathrP/var/www/html/lang_env/lib/python3.10/site-packages/pandas/tests/io/test_html.pyhtml_encoding_file(s rcOst|t|ksJdt|dt|d}ttdd||}|s(J|t||D]\}}tj||g|Ri||jrEJdq-dS)Nz*lists are not of equal size len(list1) == z, len(list2) == z$not all list elements are DataFramescSst|to t|tSN isinstancer )xyrrr>sz(assert_framelist_equal..zframes are both empty)lenallmapziptmassert_frame_equalempty)Zlist1Zlist2argskwargsmsgZ both_framesZframe_iZframe_jrrrassert_framelist_equal5s( r.cCsltd}td||ddtjtddt|ddd d dd WddS1s/wYdS) Nbs4html5lib __version__z4.2zPandas requires versionmatchrrhtml spam.htmlflavor)pytest importorskipsetattrraises ImportErrorr)Z monkeypatchrr/rrrtest_bs4_version_failsIs  "r=cCsZd}d}d|d}tjt|dtt|d|dWddS1s&wYdS)Nz google.comzinvalid flavorz\{z \} is not a valid set of flavorsr2googler3r7)r8r; ValueErrorrr)urlr7r-rrrtest_invalid_flavorRs  "rBcCsZtdtdtd|dddd}t|ddgd }t|ddgd }t||dS) Nr/lxmlr0rrr4valid_markup.htmlr) index_colr7)r8r9rr.)rfilenameZdfs_lxmlZdfs_bs4rrrtest_same_ordering[s   rGr/r0)ZmarksrCcCstt|jdS)Nr6)rrr)rrrrflavor_read_htmlfsrHc@s0eZdZddZejddZejddZddZd 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 IdddgddZWej XdddZYddZZddZ[ej jej j ddZ\ddZ]ddZ^ej IdgdddZ_ddZ`ddZaddZbddZcdS) TestReadHtmlcCs@d}tjt|d |dWddS1swYdS)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.r2a
A B
1 2
3 4
)r(Zassert_produces_warning FutureWarning)selfrHr-rrrtest_literal_html_deprecationqs"z*TestReadHtml.test_literal_html_deprecationcC|ddddS)Nrrr4r5rrKrrrr spam_datazTestReadHtml.spam_datacCrM)Nrrr4 banklist.htmlrrNrrr banklist_datarPzTestReadHtml.banklist_datacCshttjddtjtdtdddj  t }| }|t |ddid d d }t||dS) N)abcdtypecolumnsz{:.3f}classZ dataframer)attrsrE)r nprandom default_rngpdIndexlistobjectr&formatZastypefloatto_htmlrr(r))rKrHdfoutresrrrtest_to_html_compatsz TestReadHtml.test_to_html_compatc sttdtjdgddtgdddtdtjdgddtgd ddgd gd gd gd d}|dkrOttjgd tjd}ttjddtgtjd}n9|dkrqt dddl m gd }gd }nt dt gd }t gd }|j dd}td||t||dd}Wdn1swYttdtjdgddtgdddtdtjdgddtgd ddtddtgddtgd dd||d|dkrddlddl m tfddjDt|dS)NrUZInt64rW)rkrSrU?@ZFloat64)rlg@rm)TFN)TFTabc)rorpN)rorprqdefghpythonrorppyarrowr)ArrowExtensionArrayFindexzmode.string_storage dtype_backendTbooleancs$i|]}|j|ddqS)T)Z from_pandas)array).0colryexpectedparr sz3TestReadHtml.test_dtype_backend..)r r r]nanrrZobject_r r8r9Z pandas.arraysryrrfr`Zoption_contextrrxrZr(r)) rKZstring_storager}rHrgZ string_arrayZstring_array_narhresultrrrtest_dtype_backends\       zTestReadHtml.test_dtype_backendcCft|dd}|j|d||jdd}||jdd}Wdn1s'wYt||dS)Nutf-8encodingcontentFirst Federal Bank of Floridar2 Metcalf Bankopen serve_contentreadrAr.)rK httpserverrRrHrtdf1df2rrrtest_banklist_urls zTestReadHtml.test_banklist_urlcCr)Nrrr .*Water.*r2Unitr)rKrrOrHrtrrrrr test_spam_urls zTestReadHtml.test_spam_urlcCs2||dddid}||dddid}t||dS)Nz .*Florida.*idtabler3r\rr.)rKrRrHrrrrr test_banklists  zTestReadHtml.test_banklistcCsR||dd}||dd}t|||djddksJ|djddks'JdS)Nrr2rrrr ProximatesZNutrient)r.ilocrZrKrOrHrrrrr test_spam s   zTestReadHtml.test_spamcCs$||}|D] }t|tsJqdSrr)rKrOrHdfsrgrrrtest_spam_no_matchszTestReadHtml.test_spam_no_matchcCs,||ddid}|D] }t|tsJq dS)Nrr)r\r)rKrRrHrrgrrrtest_banklist_no_matchsz#TestReadHtml.test_banklist_no_matchcCs2||dddd}|jddksJ|jrJdS)NrrSr3headerrr)rZr*)rKrOrHrgrrrtest_spam_headerszTestReadHtml.test_spam_headercC*||ddd}||ddd}t||dSNrrkr3skiprowsrrrrrrtest_skiprows_int"zTestReadHtml.test_skiprows_intcC2||dtdd}||dtdd}t||dSNrrSrr)ranger.rrrrtest_skiprows_range(z TestReadHtml.test_skiprows_rangecCs2||dddgd}||dddgd}t||dSNrrkrSrrrrrrrtest_skiprows_list.rzTestReadHtml.test_skiprows_listcCs2||dddhd}||dddhd}t||dSrrrrrrtest_skiprows_set4rzTestReadHtml.test_skiprows_setcCrrrrrrrtest_skiprows_slice:rz TestReadHtml.test_skiprows_slicecCrrslicer.rrrrtest_skiprows_slice_short@rz&TestReadHtml.test_skiprows_slice_shortcCs8||dtddd}||dtdddd}t||dS) NrrSrrrTrkrrrrrtest_skiprows_slice_longFsz%TestReadHtml.test_skiprows_slice_longcCs6||dtdd}||dtdd}t||dSr)r]Zaranger.rrrrtest_skiprows_ndarrayLsz"TestReadHtml.test_skiprows_ndarraycCsBtjtdd||dddWddS1swYdS)Nz%is not a valid type for skipping rowsr2rZasdfr)r8r; TypeError)rKrOrHrrrtest_skiprows_invalidRs"z"TestReadHtml.test_skiprows_invalidcCrNrrr3rErrrrrr test_indexVszTestReadHtml.test_indexcC.||dddd}||dddd}t||dSNrrkr)r3rrErrrrrrtest_header_and_index_no_types[z+TestReadHtml.test_header_and_index_no_typescCrrrrrrr test_header_and_index_with_types`rz-TestReadHtml.test_header_and_index_with_typescCrrrrrrrtest_infer_typeseszTestReadHtml.test_infer_typescCst|dd}t|}Wdn1swYt|dd}t|}Wdn1s3wY||dd}||dd}t||dSNzUTF-8rrr2r)rrrr.)rKrOrHrtdata1data2rrrrrtest_string_ioks  zTestReadHtml.test_string_iocCsbt|dd }|}Wdn1swY|t|dd}|t|dd}t||dSr)rrrr.)rKrOrHrtrrrrrr test_stringvs  zTestReadHtml.test_stringcCs~t|dd}||dd}Wdn1swYt|dd}||dd}Wdn1s3wYt||dSr)rr.)rKrOrHrtrrrrrtest_file_likeszTestReadHtml.test_file_likecCsN|jdddtjtdd|dddWddS1s wYdS)Nz#urlopen error unknown url type: gitcoder2zgit://github.comr)rr8r;rrKrrHrrrtest_bad_url_protocols"z"TestReadHtml.test_bad_url_protocolcCsT|jdddtjttfdd||jddWddS1s#wYdS)NzName or service not knownrrzHTTP Error 404: NOT FOUNDr2r)rr8r;rr@rArrrrtest_invalid_urls"zTestReadHtml.test_invalid_urlcCsL|}|ttj|dddid}t|tsJ|D] }t|ts#JqdS)NZFirstrrr)rospathabspathr rbr rKrRrHrArrgrrr test_file_urlszTestReadHtml.test_file_urlcCsJ|}tjtdd||dddidWddS1swYdS)NzNo tables foundr2rrZ tasdfablerr8r;r@)rKrRrHrArrrtest_invalid_table_attrss  "z%TestReadHtml.test_invalid_table_attrscC0||dddiddgdd}t|jtsJdS)NMetcalfrrrrk)r3r\rr rZr rKrRrHrgrrrtest_multiindex_header z#TestReadHtml.test_multiindex_headercCr)Nrrrrrk)r3r\rE)r r{r rrrrtest_multiindex_indexrz"TestReadHtml.test_multiindex_indexcCsF||dddiddgddgdd}t|jtsJt|jts!JdS)Nrrrrrk)r3r\rrE)r rZr r{rrrrtest_multiindex_header_indexsz)TestReadHtml.test_multiindex_header_indexcC2||dddiddgddd}t|jtsJdSNrrrrrk)r3r\rrrrrrr&test_multiindex_header_skiprows_tuplesz3TestReadHtml.test_multiindex_header_skiprows_tuplescCrrrrrrrtest_multiindex_header_skiprowsrz,TestReadHtml.test_multiindex_header_skiprowscCsH||dddiddgddgddd}t|jtsJt|jts"JdS)Nrrrrrk)r3r\rrEr)r r{r rZrrrr%test_multiindex_header_index_skiprowssz2TestReadHtml.test_multiindex_header_index_skiprowscCsX|}|ttj|ttdddid}t|tsJ|D] }t|ts)Jq dS)NZFloridarrr) rrrrrecompiler rbr rrrrtest_regex_idempotencysz#TestReadHtml.test_regex_idempotencycCsFd}tjt|d||dddWddS1swYdS)Nz\(you passed a negative value\)r2ZWaterrrr)rKrOrHr-rrrtest_negative_skiprowss"z#TestReadHtml.test_negative_skiprowscCdS)Naf

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

rrKrrr python_docsszTestReadHtml.python_docscCs.|j|d||jdd}t|dksJdS)NrPythonr2rk)rrAr$)rKrrrHrrrrtest_multiple_matches(s z"TestReadHtml.test_multiple_matchescCs@|j|d||jdd}dd|D}t|ddgksJdS)Nrrr2cSsg|] }|jdddqS)rrrT)r)rrgrrr 4sz7TestReadHtml.test_python_docs_table..ZPythZWhat)rrAsorted)rKrrrHrzzrrrtest_python_docs_table/s z#TestReadHtml.test_python_docs_tablecCs$d}|t|}t|dksJdS)z@ Make sure that read_html ignores empty tables. a
A B
1 2
rkN)rr$)rKrHr4rrrrtest_empty_tables7s zTestReadHtml.test_empty_tablescCs<|tdd}tddgddggddgd }t||dS) Na
A B
1 2
3 4
rrkrSrUrTABrrZrr r(r)rKrHrrrrrtest_multiple_tbodyRsz 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
rHeaderfirstrr{Nrrrrrtest_header_and_one_columnrsz'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 rdrr(r)) rKrHZ data_templateZ expected1Z expected2rrZresult1Zresult2rrrtest_tfoot_reads   zTestReadHtml.test_tfoot_readcCs6|tdddd}tddggdd}t||dS)Na 
S I
text 1944
rrtextr)SIrYrrrrr&test_parse_header_of_non_string_columnsz3TestReadHtml.test_parse_header_of_non_string_columnc sddlmfdd}||dddidd}t|d d d d ttd d}|j|jks-Jgd}gd}||||} ||} | } ddg} | | t| | <t | | dS)Nr_remove_whitespacecs"z|WSty|YSwr)AttributeErrorr!rrr 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)) rKrRrrHrrgZ ground_trutholdnewZdfnewZgtnewZ convertedZ date_colsrrrtest_banklist_headers     z!TestReadHtml.test_banklist_headercCsnd}t|dd }|}Wdn1swY||vs"J||dddidd}||vs5JdS)Nz Gold Canyonrrrrrr)rrZ to_string)rKrRrHgcrtZraw_textrgrrrtest_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
rrEa
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))rKrHrrrrrtest_different_number_of_cols's" "$!z*TestReadHtml.test_different_number_of_colscCs6|tdd}tgdggdd}t||dS)NaZ
A B C
a b c
rrn)rrCrYrrrrrtest_colspan_rowspan_1osz#TestReadHtml.test_colspan_rowspan_1cC:|tdddd}tgdggdd}t||dS)Na
X Y Z W
A B C
rr )rrrZr$)XzX.1Yr'Wrrrrrr test_colspan_rowspan_copy_valuessz-TestReadHtml.test_colspan_rowspan_copy_valuescCr&)Na(
A B C
D
rr )rrrrD)rrzB.1zB.2r$rrrrrrtest_colspan_rowspan_both_not_1sz,TestReadHtml.test_colspan_rowspan_both_not_1cCs:|tdddd}tddggddgd}t||dS)Nz
A B
C
rr r$rrrrrrrrtest_rowspan_at_end_of_rows z'TestReadHtml.test_rowspan_at_end_of_rowcCs@|tdddd}tddgddggddgd}t||dS)Nz
A B
rr rrrrrrrrtest_rowspan_only_rowss  z#TestReadHtml.test_rowspan_only_rowscCV|tdd}tddgddggddgddggd}tdd gg|d }t||dS) Nam
A B
a b
1 2
rrrrorprkZlevelscodesrSrrr r r(r)rKrHrrZrrrr+test_header_inferred_from_rows_with_only_ths$z8TestReadHtml.test_header_inferred_from_rows_with_only_thcCshtdtdddi}|}|t|dgdd}t||d|t|dgdd}t||ddS)Ndate1/1/2001 ZperiodsrkrZ parse_datesrE)r rrfrr(r))rKrHrgrrirrrtest_parse_dates_lists z"TestReadHtml.test_parse_dates_listcCspttddd}t|dd|ddd}|t|dd d gid d }td|i}t||d dS) Nr7r8r9cS t|Sr)strr6rrrrr#% z7TestReadHtml.test_parse_dates_combine..cSr<r)r=timerrrrr#&r>)r6r?datetimerkrSr:r)r rr r&rrfr(r))rKrHZ raw_datesrgriZnewdfrrrtest_parse_dates_combine!s   z%TestReadHtml.test_parse_dates_combinecCs|dddd}tj|sJt|dtj|s%Jt|d||ddd d }|jd ks5Jd |jd vs>J|djtdksJJt |j ddsUJdS)Nrrr4wikipedia_states.htmlz is not a filez is an empty fileArizonarkrr)< Unnamedrsq mifloat64)rrGHzPN$A) rrisfilereprgetsizerrZrXr]allcloselocrKrrHrrrrrtest_wikipedia_states_table/sz(TestReadHtml.test_wikipedia_states_tablecCsn|dddd}||dddd}|jdksJd |jd d vs"J|jjd ks*Jt|jd ds5JdS)Nrrr4rBrCrr)rD rFrrkrS)ZAlaska)z Total area[2]rGrI)rrZZnlevelsr]rMrNrOrrr test_wikipedia_states_multiindex9s z-TestReadHtml.test_wikipedia_states_multiindexcCsD|tdddgd}tddggtddgd }t|d|dS) NaK
AB
ab
rrkr rorp)Unnamed: 0_level_0r)zUnnamed: 1_level_0rrY)rr r from_tuplesr(r)rrrr%test_parser_error_on_empty_header_rowAs z2TestReadHtml.test_parser_error_on_empty_header_rowcCsN|tdddd}tddidgd}|djtdksJt||dS) Na
Header
1100#101
#)decimalrrgClg0@rrH)rr rXr]r(r)rrrrtest_decimal_rowsZszTestReadHtml.test_decimal_rowsargTFcCsJtd}tjt|d|||dWddS1swYdS)NzPassing a bool to header is invalid. Use header=None for no header or header=int or list-like of ints to specify the row(s) making up the column namesr2r )rescaper8r;r)rKrOrYrHr-rrrtest_bool_header_argws "z!TestReadHtml.test_bool_header_argcCs8|tddtidd}tdddgi}t||dS)Na
a
0.763
0.244
rorrz0.763z0.244)rr=r r(r)rrrrtest_convertersszTestReadHtml.test_converterscCs8|tddgdd}tddtjgi}t||dS)Na
a
0.763
0.244
gZd;?)Z na_valuesrrog"~j?rr r]rr(r)rrrrtest_na_valuesszTestReadHtml.test_na_valuescCsld}tdddgi}|t|ddd}t||tdtjtjgi}|t|ddd}t||dS) Na
a
N/A
NA
rozN/Ar F)Zkeep_default_narT)r rr(r)r]r)rKrHZ html_data expected_dfhtml_dfrrrtest_keep_default_nas z!TestReadHtml.test_keep_default_nacCs@|tdd}tddgtjtjggddgd}t||dS)Nak
A B
a b
rrorprrrr]rrrrtest_preserve_empty_rowss z%TestReadHtml.test_preserve_empty_rowscCr0) NaU
AB
ab
12
rrrrorprkr1rSrr3r4rrr,test_ignore_empty_rows_when_inferring_headers$z9TestReadHtml.test_ignore_empty_rows_when_inferring_headercCsLtgdd}gdgdg|_|jdd}|t|d}t||dS)N))ZHillaryDr,)ZBernieJr,)ZDonaldER)r)rSZAgeZParty)NamezUnnamed: 1_level_1zUnnamed: 2_level_1Frzr)r rZrfrr(r))rKrHr_r4r`rrrtest_multiple_header_rowss z&TestReadHtml.test_multiple_header_rowscCs>|dddd}||dd}t|tsJt|dtsJdS)Nrrr4rDrr")r rbr )rKrrHrFrrrrtest_works_on_valid_markups z'TestReadHtml.test_works_on_valid_markupcCs$|dddd}||dddgddS) Nrrr4rQrrCr0r?r)rKrrHrRrrrtest_fallback_successsz"TestReadHtml.test_fallback_successcCs>tddd}ttjdd|d}|}d|vsJdS)Nz 2000-01-01r8r9rS)r8rTrz)rr r]r^r_Zstandard_normalrf)rKrngrgrrrrtest_to_html_timestamps z#TestReadHtml.test_to_html_timestampcCstdddg}|}|jdd}|jdd}|jdd}|jdd}|jdd}d|vs0J||ks6J||ks
foo bar baz qux
foo
displayed_onlyrrk)rr(r)r$)rKrrZexp0Zexp1rHrrrrrtest_displayed_only5s  z TestReadHtml.test_displayed_onlyrrcCs>d}|t||dd}tddgddgd}t||dS) NaW
A B
1 2
4 5
rqrrkrTrSrrnr)rKrrrHZ html_tablerrrrr&test_displayed_only_with_many_elementsZs 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\}}zZt|d}|||dd}Wdn1s1wYt|d}|t||dd} Wdn1sSwY|||dd} t || t || WdSt yt rd|vsd|vrt w)Nr_rb)rrEZ16Z32)rrbasenamesplitextsplitrrpoprr(r) Exceptionrr8skip) rKrrH base_pathrootrurZfobjZ from_stringZfrom_file_like from_filenamerrr test_encoders>      zTestReadHtml.test_encodecCsz|jddkr tdGdddt}|d}||sJtjtdd ||WddS1s6wYdS) Nr7rCzNot applicable for lxmlc@seZdZddZdS)zFTestReadHtml.test_parse_failure_unseekable..UnseekableStringIOcSrNFrrrrrseekablezOTestReadHtml.test_parse_failure_unseekable..UnseekableStringIO.seekableN)__name__ __module__ __qualname__rrrrrUnseekableStringIOs rz?
spameggs
z#passed a non-rewindable file objectr2)keywordsgetr8r|rr;r@)rKrHrbadrrrtest_parse_failure_unseekables   "z*TestReadHtml.test_parse_failure_unseekablecCs:Gddd}|d}|d}||sJ||sJdS)Nc@sFeZdZdddZdddZddZd d Zd d Zdefd dZ dS)z9TestReadHtml.test_parse_failure_rewinds..MockFilereturnNcSs||_d|_dSr)rat_end)rKrrrr__init__s zBTestReadHtml.test_parse_failure_rewinds..MockFile.__init__cSs|jrdn|j}d|_|S)NrT)rr)rKsizerrrrrsz>TestReadHtml.test_parse_failure_rewinds..MockFile.readcSs d|_dSr)r)rKoffsetrrrseeks z>TestReadHtml.test_parse_failure_rewinds..MockFile.seekcSr)NTrrrrrrrzBTestReadHtml.test_parse_failure_rewinds..MockFile.seekablecSsdSrrrrrr__next__rzBTestReadHtml.test_parse_failure_rewinds..MockFile.__next__cSs|Srrrrrr__iter__szBTestReadHtml.test_parse_failure_rewinds..MockFile.__iter__)rNr) rrrrrrrrrrrrrrMockFiles  rz/
spam
eggs
z2
spameggs
r)rKrHrZgoodrrrrtest_parse_failure_rewindss  z'TestReadHtml.test_parse_failure_rewindscCsGdddtj}|dddd}|||fd}|||fd}|||s.|r7 |s.|s.d|jurD|jusGJJdS)NcseZdZfddZZS)z@TestReadHtml.test_importcheck_thread_safety..ErrorThreadc sFztWnty}z ||_WYd}~dSd}~wwd|_dSr)superrunr{err)rKr __class__rrrs zDTestReadHtml.test_importcheck_thread_safety..ErrorThread.run)rrrr __classcell__rrrr ErrorThreadsrrrr4rD)targetr+) threadingThreadstartis_aliver)rKrrHrrFZhelper_thread1Zhelper_thread2rrrtest_importcheck_thread_safetys $z+TestReadHtml.test_importcheck_thread_safetycCs>|dddd}t|}||d}||d}t||dS)Nrrr4r5r)rr(r))rKrrHZfile_path_string file_pathrrrrrtest_parse_path_objects   z#TestReadHtml.test_parse_path_objectcCs2|tdd}tdggdgd}t||dS)Nz
A
word1
word2
rz word1 word2rrrrrrrtest_parse_br_as_spaces z#TestReadHtml.test_parse_br_as_space)r%bodyrr c Csd}gdgdgdgdgdgdd}|d }|d }|d }|d kr4|d }|d}|d}n|dkr=|d }n|dkrF|d}n|dkrN|d}|t||dd}t||g|d} | tj} t|| dS)Na
HTTP FTP Linkless
Wikipedia SURROUNDING Debian TEXT Linkless
Footer Multiple links: Only first captured.
)HTTPFTPLinkless))rN)rN)rz'https://en.wiktionary.org/wiki/linkless) WikipediaSURROUNDING Debian TEXTr))rzhttps://en.wikipedia.org/)rzftp://ftp.us.debian.org/)rN)Footer$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%rrrrr rZ extract_linksrrY)rr Zfillnar]rr(r)) rKrYrHZ gh_13141_dataZgh_13141_expectedZdata_expZfoot_expZhead_exprrrrrtest_extract_linkss2    zTestReadHtml.test_extract_linkscCsDd}tjt|dt|ddWddS1swYdS)NzY`extract_links` must be one of {None, "header", "footer", "body", "all"}, got "incorrect"r2Z incorrectrr8r;r@r)rKrOr-rrrtest_extract_links_badC "z#TestReadHtml.test_extract_links_badcCs4d}|t|ddd}tdgg}t||dS)Nz
Google.com
r%rr)z Google.comzhttps://google.comrrKrHrrrrrr test_extract_links_all_no_headerKs z-TestReadHtml.test_extract_links_all_no_headercCsDd}tjt|dtdddWddS1swYdS)NzPdtype_backend numpy is invalid, only 'numpy_nullable' and 'pyarrow' are allowed.r2testnumpyr|r)rKr-rrrtest_invalid_dtype_backendZrz'TestReadHtml.test_invalid_dtype_backendcCs@d}|t|d}tddgddggddgd }t||dS) Na
A B
A1 B1
A2 B2
rA1ZB1A2ZB2rrrrrrrrtest_style_tagbszTestReadHtml.test_style_tag)drrrrLr8fixturerOrRrjrmarknetworkZ single_cpurrZslowrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr rrr!r#r%r+r-r.r/r5r;rArPrRrUrXZ parametrizer[r\r^rarbrcrirjrkrmrpr rsrtfilterwarningsrrrrrrrrrrrrrrrrIps  <                   0    & /  H#         "  D rI)4collections.abcr functoolsrrrrrpathlibrrr urllib.errorrrr]r8Z pandas.compatrZpandas.util._test_decoratorsutilZ_test_decoratorstdZpandasr`r r r r r rrrrZpandas._testingZ_testingr(Zpandas.core.arraysrrZpandas.io.commonrrrr.r=rBrGrZ skip_if_norHrIrrrrs@    ,