o vh@s^ddlmZddlmZddlmZmZddlZddlm Z ddl Z ddl Z ddl m Z ddlZddlZddlmZddlmmZddlZddlmZmZmZmZmZmZmZm Z m!Z!ddl"m#Z$dd l%m&Z&ej'gd d d d Z(ddZ)ddZ*ddZ+ddZ,ej'ej-de.de.dgdej-de.ddgd ddZ/GdddZ0dS))Iterator)partial)BytesIOStringION)Path)URLError)is_platform_windows) NA DataFrame MultiIndexSeries Timestamp date_rangeread_csv read_html to_datetime)file_path_to_url)zchinese_utf-16.htmlzchinese_utf-32.htmlzchinese_utf-8.htmlzletz_latin1.html)paramscCs|ddd|jS)z6Parametrized fixture for HTML encoding test filenames.iodata html_encoding)param)requestdatapathrW/var/www/html/hyperkenya/venv/lib/python3.10/site-packages/pandas/tests/io/test_html.pyhtml_encoding_file$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$not all list elements are DataFramescSst|to t|tSN isinstancer )xyrrr:sz(assert_framelist_equal..zframes are both empty)lenallmapziptmassert_frame_equalempty)list1list2argskwargsmsg both_framesframe_iframe_jrrrassert_framelist_equal1s( r2cCsltd}td||ddtjtddt|ddd d dd WddS1s/wYdS) Nbs4html5lib __version__z4.2zPandas requires versionmatchrrhtml spam.htmlflavor)pytest importorskipsetattrraises ImportErrorr) monkeypatchrr3rrrtest_bs4_version_failsEs  "rBcCsZd}d}d|d}tjt|dtt|d|dWddS1s&wYdS)Nz google.comzinvalid flavorz\{z \} is not a valid set of flavorsr6googler7r;)r<r? ValueErrorrr)urlr;r.rrrtest_invalid_flavorNs  "rGcCsZtdtdtd|dddd}t|ddgd }t|ddgd }t||dS) Nr3lxmlr4rrr8valid_markup.htmlr) index_colr;)r<r=rr2)rfilenamedfs_lxmldfs_bs4rrrtest_same_orderingWs   rNr3r4)marksrHcCstt|jdS)Nr:)rrr)rrrrflavor_read_htmlbsrPc@s8eZdZddZejddZejddZddZd d Z ej j ej j d d Z ej j ej j d dZej jddZddZddZddZddZddZddZddZdd Zd!d"Zd#d$Zd%d&Zd'd(Zd)d*Zd+d,Zd-d.Zd/d0Z d1d2Z!d3d4Z"d5d6Z#d7d8Z$ej j ej j d9d:Z%ej jej j ej j d;d<Z&ej jd=d>Z'ej jd?d@Z(ej jdAdBZ)ej jdCdDZ*ej jdEdFZ+ej jdGdHZ,ej jdIdJZ-ej jdKdLZ.ej jdMdNZ/dOdPZ0ejdQdRZ1ej j ej j dSdTZ2ej j ej j dUdVZ3dWdXZ4dYdZZ5d[d\Z6d]d^Z7d_d`Z8dadbZ9ej jdcddZ:ej jdedfZ;dgdhZdmdnZ?dodpZ@dqdrZAdsdtZBdudvZCdwdxZDdydzZEd{d|ZFd}d~ZGddZHej IdddgddZJddZKddZLddZMddZNddZOddZPddZQej jddZRddZSddZTej IddeUdgdfdeUdgeUdgfgddZVej IdddgddZWeXYej ZdddZ[ddZ\ddZ]ej jej j ddZ^ddZ_ddZ`ej IdgdddZaddZbddZcddZdddZedS) TestReadHtmlcCs@d}tjt|d |dWddS1swYdS)NzPassing literal html to 'read_html' is deprecated and will be removed in a future version. To read from a literal string, wrap it in a 'StringIO' object.r6a
A B
1 2
3 4
)r'assert_produces_warning FutureWarning)selfrPr.rrrtest_literal_html_deprecationms"z*TestReadHtml.test_literal_html_deprecationcC|ddddS)Nrrr8r9rrTrrrr spam_datazTestReadHtml.spam_datacCrV)Nrrr8 banklist.htmlrrWrrr banklist_datarYzTestReadHtml.banklist_datacCsdttjddttdddj t }| }|t |ddidd d}t ||dS) N)abccolumnsz{:.3f}class dataframer)attrsrJ)r nprandom default_rngpdIndexlistr%formatastypefloatto_htmlrr'r()rTrPdfoutresrrrtest_to_html_compats z TestReadHtml.test_to_html_compatc sttdtjdgddtgdddtdtjdgddtgd ddgd gd gd gd d}|jdd}td||t||dd}Wdn1sRwY|dkrht dt  }nt |}ttdtjdgddtgdddtdtjdgddtgd ddtddt gddtgd ddtgd |dtgd |dd|dkrddlddlmtfddjDtj|dddS)Nr^Int64)dtype)rsr\r^?@Float64)rvg@rw)TFN)TFTabc)rzr{N)rzr{r|defghFindexzmode.string_storage dtype_backendrpyarrowTboolean)ArrowExtensionArraycs$i|]}|j|ddqS)T) from_pandas)array).0colrexpectedparr sz3TestReadHtml.test_dtype_backend..)check_column_type)r r renanrnrhoption_contextrr<r= ArrowDtypestring StringDtyper r pandas.arraysrrar'r()rTstring_storagerrProrpresult string_dtyperrrtest_dtype_backendsN      zTestReadHtml.test_dtype_backendcCft|dd}|j|d||jdd}||jdd}Wdn1s'wYt||dS)Nutf-8encodingcontentFirst Federal Bank of Floridar6 Metcalf Bankopen serve_contentreadrFr2)rT httpserverr[rPrdf1df2rrrtest_banklist_urls zTestReadHtml.test_banklist_urlcCr)Nrrr .*Water.*r6Unitr)rTrrXrPrrrrrr test_spam_urls zTestReadHtml.test_spam_urlcCs2||dddid}||dddid}t||dS)Nz .*Florida.*idtabler7rdrr2)rTr[rPrrrrr test_banklists  zTestReadHtml.test_banklistcCsR||dd}||dd}t|||djddksJ|djddks'JdS)Nrr6rrrr ProximatesNutrient)r2ilocrarTrXrPrrrrr test_spams   zTestReadHtml.test_spamcCs$||}|D] }t|tsJqdSrr)rTrXrPdfsrorrrtest_spam_no_match szTestReadHtml.test_spam_no_matchcCs,||ddid}|D] }t|tsJq dS)Nrr)rdr)rTr[rPrrorrrtest_banklist_no_matchsz#TestReadHtml.test_banklist_no_matchcCs2||dddd}|jddksJ|jrJdS)Nrr\r7headerrr)rar))rTrXrProrrrtest_spam_headerszTestReadHtml.test_spam_headercC*||ddd}||ddd}t||dSNrrsr7skiprowsrrrrrrtest_skiprows_intzTestReadHtml.test_skiprows_intcC2||dtdd}||dtdd}t||dSNrr\rr)ranger2rrrrtest_skiprows_rangez TestReadHtml.test_skiprows_rangecCs2||dddgd}||dddgd}t||dSNrrsr\rrrrrrrtest_skiprows_list$rzTestReadHtml.test_skiprows_listcCs2||dddhd}||dddhd}t||dSrrrrrrtest_skiprows_set*rzTestReadHtml.test_skiprows_setcCrrrrrrrtest_skiprows_slice0rz TestReadHtml.test_skiprows_slicecCrrslicer2rrrrtest_skiprows_slice_short6rz&TestReadHtml.test_skiprows_slice_shortcCs8||dtddd}||dtdddd}t||dS) Nrr\rrr]rsrrrrrtest_skiprows_slice_long<sz%TestReadHtml.test_skiprows_slice_longcCs6||dtdd}||dtdd}t||dSr)rearanger2rrrrtest_skiprows_ndarrayBsz"TestReadHtml.test_skiprows_ndarraycCsBtjtdd||dddWddS1swYdS)Nz%is not a valid type for skipping rowsr6rasdfr)r<r? TypeError)rTrXrPrrrtest_skiprows_invalidHs"z"TestReadHtml.test_skiprows_invalidcCrNrrr7rJrrrrrr test_indexLszTestReadHtml.test_indexcC.||dddd}||dddd}t||dSNrrsr)r7rrJrrrrrrtest_header_and_index_no_typesQz+TestReadHtml.test_header_and_index_no_typescCrrrrrrr test_header_and_index_with_typesVrz-TestReadHtml.test_header_and_index_with_typescCrrrrrrrtest_infer_types[szTestReadHtml.test_infer_typescCst|dd}t|}Wdn1swYt|dd}t|}Wdn1s3wY||dd}||dd}t||dSNzUTF-8rrr6r)rrrr2)rTrXrPrdata1data2rrrrrtest_string_ioas  zTestReadHtml.test_string_iocCsbt|dd }|}Wdn1swY|t|dd}|t|dd}t||dSr)rrrr2)rTrXrPrrrrrrr test_stringls  zTestReadHtml.test_stringcCs~t|dd}||dd}Wdn1swYt|dd}||dd}Wdn1s3wYt||dSr)rr2)rTrXrPrrrrrrtest_file_likeuszTestReadHtml.test_file_likecCsN|jdddtjtdd|dddWddS1s wYdS)Nz#urlopen error unknown url type: gitcoder6zgit://github.comr)rr<r?rrTrrPrrrtest_bad_url_protocol~s"z"TestReadHtml.test_bad_url_protocolcCsT|jdddtjttfdd||jddWddS1s#wYdS)NzName or service not knownrrzHTTP Error 404: NOT FOUNDr6r)rr<r?rrErFrrrrtest_invalid_urls"zTestReadHtml.test_invalid_urlcCsL|}|ttj|dddid}t|tsJ|D] }t|ts#JqdS)NFirstrrr)rospathabspathrrjr rTr[rPrFrrorrr test_file_urlszTestReadHtml.test_file_urlcCsJ|}tjtdd||dddidWddS1swYdS)NzNo tables foundr6rr tasdfablerr<r?rE)rTr[rPrFrrrtest_invalid_table_attrss  "z%TestReadHtml.test_invalid_table_attrscC0||dddiddgdd}t|jtsJdS)NMetcalfrrrrs)r7rdrrrar rTr[rProrrrtest_multiindex_header z#TestReadHtml.test_multiindex_headercCr)Nrrrrrs)r7rdrJ)rrr rrrrtest_multiindex_indexrz"TestReadHtml.test_multiindex_indexcCsF||dddiddgddgdd}t|jtsJt|jts!JdS)Nrrrrrs)r7rdrrJ)rrar rrrrrtest_multiindex_header_indexsz)TestReadHtml.test_multiindex_header_indexcC2||dddiddgddd}t|jtsJdSNrrrrrs)r7rdrrrrrrr&test_multiindex_header_skiprows_tuplesz3TestReadHtml.test_multiindex_header_skiprows_tuplescCrrrrrrrtest_multiindex_header_skiprowsrz,TestReadHtml.test_multiindex_header_skiprowscCsH||dddiddgddgddd}t|jtsJt|jts"JdS)Nrrrrrs)r7rdrrJr)rrr rarrrr%test_multiindex_header_index_skiprowssz2TestReadHtml.test_multiindex_header_index_skiprowscCsX|}|ttj|ttdddid}t|tsJ|D] }t|ts)Jq dS)NFloridarrr) rrrrrecompilerrjr rrrrtest_regex_idempotencysz#TestReadHtml.test_regex_idempotencycCsFd}tjt|d||dddWddS1swYdS)Nz\(you passed a negative value\)r6Waterrrr)rTrXrPr.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

rrTrrr python_docsszTestReadHtml.python_docscCs.|j|d||jdd}t|dksJdS)NrPythonr6rs)rrFr#)rTrrrPrrrrtest_multiple_matchess z"TestReadHtml.test_multiple_matchescCs@|j|d||jdd}dd|D}t|ddgksJdS)Nrrr6cSsg|] }|jdddqS)rrr])r)rrorrr *sz7TestReadHtml.test_python_docs_table..PythWhat)rrFsorted)rTrrrPrzzrrrtest_python_docs_table%s z#TestReadHtml.test_python_docs_tablecCs$d}|t|}t|dksJdS)z@ Make sure that read_html ignores empty tables. a
A B
1 2
rsN)rr#)rTrPr8rrrrtest_empty_tables-s zTestReadHtml.test_empty_tablescCs<|tdd}tddgddggddgd }t||dS) Na
A B
1 2
3 4
rrsr\r^r]ABrrarr r'r(rTrPrrrrrtest_multiple_tbodyHsz TestReadHtml.test_multiple_tbodycCs2|tdd}tddidgd}t||dS)zt Don't fail with bs4 when there is a header and only one column as described in issue #9178 a3
Header
first
rHeaderfirstrrNrrrrrtest_header_and_one_columnhsz'TestReadHtml.test_header_and_one_columncCs6|tdd}tgdggdd}t||dS)zK Ensure parser adds within on malformed HTML. a
Country Municipality Year
Ukraine Odessa 1944
r)UkraineOdessa)Country MunicipalityYearrNrrrrrtest_thead_without_trsz"TestReadHtml.test_thead_without_trc Csd}tddggddgd}tddgddggddgd}|jd d }|jd d }|t|d }|t|d }t||t||d S)zh Make sure that read_html reads tfoot, containing td or th. Ignores empty tfoot a {footer}
A B
bodyA bodyB
bodyAbodyBrrrfootAfootB)footerz%footAfootBrN)r rkrr'r() rTrP data_template expected1 expected2rrresult1result2rrrtest_tfoot_reads   zTestReadHtml.test_tfoot_readcCs6|tdddd}tddggdd}t||dS)Na 
S I
text 1944
rrtextr$)SIr`rrrrr&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 r:rr 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 RicoEurobankzSanderson State BankzWashington Mutual BankzSilver State BankzAmTrade International BankzHamilton Bank, NAzThe Citizens Savings BankrAr@) pandas.io.htmlr;rr shaper%replaceapplyrr'r() rTr[rrPr>ro ground_trutholdnewdfnewgtnew converted date_colsrr:rtest_banklist_headers     z!TestReadHtml.test_banklist_headercCsnd}t|dd }|}Wdn1swY||vs"J||dddidd}||vs5JdS)Nz Gold Canyonrrrrrr)rr to_string)rTr[rPgcrraw_textrorrrtest_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
rrJa
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()rTrPrrrrrtest_different_number_of_colss" "$!z*TestReadHtml.test_different_number_of_colscCs6|tdd}tgdggdd}t||dS)NaZ
A B C
a b c
rry)rrCr`rrrrrtest_colspan_rowspan_1esz#TestReadHtml.test_colspan_rowspan_1cC:|tdddd}tgdggdd}t||dS)Na
X Y Z W
A B C
rr5)rrrZrW)XzX.1YrZWrrrrrr test_colspan_rowspan_copy_values~sz-TestReadHtml.test_colspan_rowspan_copy_valuescCrY)Na(
A B C
D
rr5)rrrrD)rrzB.1zB.2rWrrrrrrtest_colspan_rowspan_both_not_1sz,TestReadHtml.test_colspan_rowspan_both_not_1cCs:|tdddd}tddggddgd}t||dS)Nz
A B
C
rr5rWrrrrrrrrtest_rowspan_at_end_of_rows z'TestReadHtml.test_rowspan_at_end_of_rowcCs@|tdddd}tddgddggddgd}t||dS)Nz
A B
rr5rrrrrrrrtest_rowspan_only_rowss  z#TestReadHtml.test_rowspan_only_rowscCV|tdd}tddgddggddgddggd}tdd gg|d }t||dS) Nam
A B
a b
1 2
rrrrzr{rslevelscodesr\rrr r r'r(rTrPrrarrrr+test_header_inferred_from_rows_with_only_ths$z8TestReadHtml.test_header_inferred_from_rows_with_only_thcCshtdtdddi}|}|t|dgdd}t||d|t|dgdd}t||ddS)Ndate1/1/2001 periodsrsr parse_datesrJ)r rrnrr'r()rTrProrrqrrrtest_parse_dates_lists z"TestReadHtml.test_parse_dates_listcCspttddd}t|dd|ddd}|t|dd d gid d }td|i}t||d dS) NrkrlrmcS t|Sr)strrjr=rrrr" z7TestReadHtml.test_parse_dates_combine..cSrrr)rstimer=rrrr"rt)rjrudatetimersr\ror)r rr r%rrnr'r()rTrP raw_datesrorqnewdfrrrtest_parse_dates_combines   z%TestReadHtml.test_parse_dates_combinecCs|dddd}tj|sJt|dtj|s%Jt|d||ddd d }|jd ks5Jd |jd vs>J|djtdksJJt |j ddsUJdS)Nrrr8wikipedia_states.htmlz is not a filez is an empty fileArizonarsrr)< Unnamedrsq mifloat64)rrHzPN$A) rrisfilereprgetsizerFrarureallcloselocrTrrPrrrrrtest_wikipedia_states_table%sz(TestReadHtml.test_wikipedia_states_tablecCsn|dddd}||dddd}|jdksJd |jd d vs"J|jjd ks*Jt|jd ds5JdS)Nrrr8rzr{rr)r| r~rrsr\)Alaska)z Total area[2]rr)rFranlevelsrerrrrrr test_wikipedia_states_multiindex/s z-TestReadHtml.test_wikipedia_states_multiindexcCsD|tdddgd}tddggtddgd }t|d|dS) NaK
AB
ab
rrsr5rzr{)Unnamed: 0_level_0r)zUnnamed: 1_level_0rr`)rr r from_tuplesr'r(rrrr%test_parser_error_on_empty_header_row7s z2TestReadHtml.test_parser_error_on_empty_header_rowcCsN|tdddd}tddidgd}|djtdksJt||dS) Na
Header
1100#101
#)decimalrrgClg0@r r)rr rurer'r(rrrrtest_decimal_rowsPszTestReadHtml.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 namesr6r5)rescaper<r?r)rTrXrrPr.rrrtest_bool_header_argms "z!TestReadHtml.test_bool_header_argcCs8|tddtidd}tdddgi}t||dS)Na
a
0.763
0.244
rzrBrz0.763z0.244)rrsr r'r(rrrrtest_convertersxszTestReadHtml.test_converterscCs8|tddgdd}tddtjgi}t||dS)Na
a
0.763
0.244
gZd;?) na_valuesrrzg"~j?rr rerr'r(rrrrtest_na_valuesszTestReadHtml.test_na_valuescCsld}tdddgi}|t|ddd}t||tdtjtjgi}|t|ddd}t||dS) Na
a
N/A
NA
rzzN/Ar F)keep_default_narT)r rr'r(rer)rTrP html_data expected_dfhtml_dfrrrtest_keep_default_nas z!TestReadHtml.test_keep_default_nacCs@|tdd}tddgtjtjggddgd}t||dS)Nak
A B
a b
rrzr{rrrrrrrrtest_preserve_empty_rowss z%TestReadHtml.test_preserve_empty_rowscCrc) NaU
AB
ab
12
rrrrzr{rsrdr\rrgrhrrr,test_ignore_empty_rows_when_inferring_headers$z9TestReadHtml.test_ignore_empty_rows_when_inferring_headercCsLtgdd}gdgdg|_|jdd}|t|d}t||dS)N))HillaryDr_)BernieJr_)DonaldER)r)rAgeParty)NamezUnnamed: 1_level_1zUnnamed: 2_level_1Frr)r rarnrr'r()rTrPrr8rrrrtest_multiple_header_rowss z&TestReadHtml.test_multiple_header_rowscCs>|dddd}||dd}t|tsJt|dtsJdS)Nrrr8rIrrU)rrjr )rTrrPrKrrrrtest_works_on_valid_markups z'TestReadHtml.test_works_on_valid_markupcCs$|dddd}||dddgddS) Nrrr8rZrrHr4rDr)rTrrPr[rrrtest_fallback_success sz"TestReadHtml.test_fallback_successcCs>tddd}ttjdd|d}|}d|vsJdS)Nz 2000-01-01rlrmr\)rlr]r)rr rerfrgstandard_normalrn)rTrngrorrrrtest_to_html_timestamps z#TestReadHtml.test_to_html_timestampcCstdddg}|}|jdd}|jdd}|jdd}|jdd}|jdd}d|vs0J||ks6J||ks
foo bar baz qux
foo
displayed_onlyrrs)rr'r(r#)rTrexp0exp1rPrrrrrtest_displayed_only+s  z TestReadHtml.test_displayed_onlyrcCs>d}|t||dd}tddgddgd}t||dS) NaW
A B
1 2
4 5
rrrsr]r\rrr)rTrrP html_tablerrrrr&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\}}zZt|d}|||dd}Wdn1s1wYt|d}|t||dd} Wdn1sSwY|||dd} t || t || WdSt yt rd|vsd|vrt w)Nr_rb)rrJ1632)rrbasenamesplitextsplitrrpoprr'r( Exceptionrr<skip) rTrrP base_pathrootrrfobj from_stringfrom_file_like from_filenamerrr test_encodehs>      zTestReadHtml.test_encodecCsz|jddkr tdGdddt}|d}||sJtjtdd ||WddS1s6wYdS) Nr;rHzNot applicable for lxmlc@seZdZddZdS)zFTestReadHtml.test_parse_failure_unseekable..UnseekableStringIOcSr NFrr rrrseekablezOTestReadHtml.test_parse_failure_unseekable..UnseekableStringIO.seekableN)__name__ __module__ __qualname__rrrrrUnseekableStringIOs rz?
spameggs
z#passed a non-rewindable file objectr6)keywordsgetr<rrr?rE)rTrPrbadrrrtest_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)rTrrrr__init__s zBTestReadHtml.test_parse_failure_rewinds..MockFile.__init__cSs|jrdn|j}d|_|S)Nr-T)rr)rTsizerrrrrsz>TestReadHtml.test_parse_failure_rewinds..MockFile.readcSs d|_dSr)r)rToffsetrrrseeks z>TestReadHtml.test_parse_failure_rewinds..MockFile.seekcSr )NTrr rrrrrzBTestReadHtml.test_parse_failure_rewinds..MockFile.seekablecSsdSrrr rrr__next__rzBTestReadHtml.test_parse_failure_rewinds..MockFile.__next__cSs|Srrr rrr__iter__szBTestReadHtml.test_parse_failure_rewinds..MockFile.__iter__)rNr) rrrrrrrrrrrrrrMockFiles  rz/
spam
eggs
z2
spameggs
r)rTrPrgoodrrrrtest_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)superrunrerr)rTr __class__rrrs zDTestReadHtml.test_importcheck_thread_safety..ErrorThread.run)rrrr __classcell__rrrr ErrorThreadsrrrr8rI)targetr,) threadingThreadstartis_aliver)rTrrPrrKhelper_thread1helper_thread2rrrtest_importcheck_thread_safetys $z+TestReadHtml.test_importcheck_thread_safetycCs>|dddd}t|}||d}||d}t||dS)Nrrr8r9r)rr'r()rTrrPfile_path_string file_pathrrrrrtest_parse_path_objects   z#TestReadHtml.test_parse_path_objectcCs2|tdd}tdggdgd}t||dS)Nz
A
word1
word2
rz word1 word2rrrrrrrtest_parse_br_as_spaces z#TestReadHtml.test_parse_br_as_space)r$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$Multiple links: Only first captured.N))rz)https://en.wikipedia.org/wiki/Page_footer)r1N) head_ignore head_extract body_ignore body_extract footer_ignorefooter_extractrr rr$rr rrr.r extract_linksrr`)rr fillnarerr'r() rTrrP gh_13141_datagh_13141_expecteddata_expfoot_exphead_exprrrrrtest_extract_linkss2    zTestReadHtml.test_extract_linkscCsDd}tjt|dt|ddWddS1swYdS)NzY`extract_links` must be one of {None, "header", "footer", "body", "all"}, got "incorrect"r6 incorrectr r<r?rEr)rTrXr.rrrtest_extract_links_bad: "z#TestReadHtml.test_extract_links_badcCs4d}|t|ddd}tdgg}t||dS)Nz
Google.com
r$r r)z Google.comzhttps://google.comrrTrPrrrrrr test_extract_links_all_no_headerBs z-TestReadHtml.test_extract_links_all_no_headercCsDd}tjt|dtdddWddS1swYdS)NzPdtype_backend numpy is invalid, only 'numpy_nullable' and 'pyarrow' are allowed.r6testnumpyrr)rTr.rrrtest_invalid_dtype_backendQrz'TestReadHtml.test_invalid_dtype_backendcCs@d}|t|d}tddgddggddgd }t||dS) Na
A B
A1 B1
A2 B2
rA1B1A2B2rrrrrrrrtest_style_tagYszTestReadHtml.test_style_tag)frrrrUr<fixturerXr[rrrmarknetwork single_cpurrslowrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr r rrrrrr!r(r4r9rPrTrVrXr^r`rarbrirqryrrrr parametrizerrrrrrrrrrrr rrtdskip_if_windowsfilterwarningsrrrrrrrrrrr!rrrrrQls   6                   0    & /  H#         "  D rQ)1collections.abcr functoolsrrrrrpathlibrrr urllib.errorrrrer< pandas.compatrpandas.util._test_decoratorsutil_test_decoratorsr(pandasrhr r r r r rrrrpandas._testing_testingr'pandas.io.commonrr"rr2rBrGrNr skip_if_norPrQrrrrs>    ,