o Ti @shddlmZddlmZddlmZmZddlZddlm Z ddl Z ddl Z ddl m Z ddlZddlZddlmZddlmmZddlZddlmZmZmZmZmZmZmZm Z m!Z!ddl"m#Z$dd l%m&Z&m'Z'dd l(m)Z)ej*gd d d dZ+ddZ,ddZ-ddZ.ddZ/ej01dej2de3de3dgdej2de3ddgGdddZ4dS))Iterator)partial)BytesIOStringION)Path)URLError)is_platform_windows) NA DataFrame MultiIndexSeries Timestamp date_rangeread_csv read_html to_datetime)ArrowStringArray StringArray)file_path_to_url)zchinese_utf-16.htmlzchinese_utf-32.htmlzchinese_utf-8.htmlzletz_latin1.html)paramscCs|ddd|jS)z6Parametrized fixture for HTML encoding test filenames.iodataZ html_encoding)param)requestdatapathrY/var/www/html/evchargy.com/venv/lib/python3.10/site-packages/pandas/tests/io/test_html.pyhtml_encoding_file(s rcOst|t|ksJdt|dt|d}ttdd||}|s(J|t||D]\}}tj||g|Ri||jrEJdq-dS)Nz*lists are not of equal size len(list1) == z, len(list2) == z$not all list elements are DataFramescSst|to t|tSN) isinstancer )xyrrr>sz(assert_framelist_equal..zframes are both empty)lenallmapziptmassert_frame_equalempty)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 flavorsr1Zgoogler2r6)r7r: ValueErrorrr)urlr6r,rrrtest_invalid_flavorRs  "r@cCsZtdtdtd|dddd}t|ddgd }t|ddgd }t||dS) Nr.lxmlr/rrr3valid_markup.htmlr) index_colr6)r7r8rr-)rfilenameZdfs_lxmlZdfs_bs4rrrtest_same_ordering[s   rEr6r.r/)ZmarksrAc@sLeZdZddZejddZejddZejddd d Zd d Z d dZ ej j ej j ddZej j ej j ddZej jddZddZddZddZddZddZdd Zd!d"Zd#d$Zd%d&Zd'd(Zd)d*Zd+d,Zd-d.Zd/d0Zd1d2Z d3d4Z!d5d6Z"d7d8Z#d9d:Z$d;d<Z%ej j ej j d=d>Z&ej jej j ej j d?d@Z'ej jdAdBZ(ej jdCdDZ)dEdFZ*ej jdGdHZ+ej jdIdJZ,ej jdKdLZ-ej jdMdNZ.ej jdOdPZ/ej jdQdRZ0ej jdSdTZ1dUdVZ2ejdWdXZ3ej j ej j dYdZZ4ej j ej j d[d\Z5d]d^Z6d_d`Z7dadbZ8dcddZ9dedfZ:dgdhZ;ej jdidjZdodpZ?dqdrZ@dsdtZAdudvZBdwdxZCdydzZDd{d|ZEd}d~ZFddZGddZHddZIddZJej KdddgddZLddZMddZNddZOddZPddZQddZRddZSej jddZTddZUddZVej KddeWdgdfdeWdgeWdgfgddZXej KdddgddZYej ZdddZ[ddZ\ddZ]ej jej j ddZ^ddZ_ddZ`ej KdgdddZaddZbddZcddZdddZedS) TestReadHtmlcCsBd}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.r1a
A B
1 2
3 4
)r'Zassert_produces_warning FutureWarningrselfr,rrrtest_literal_html_deprecationns"z*TestReadHtml.test_literal_html_deprecationcC|ddddS)Nrrr3r4rrIrrrr spam_datazTestReadHtml.spam_datacCrK)Nrrr3 banklist.htmlrrLrrr banklist_datarNzTestReadHtml.banklist_dataT)Zautouseccstt|d|_dVdS)Nr5)rr)rIr6rrr set_defaultss zTestReadHtml.set_defaultscCsZtjddddddddjt}|}|jt|dd id d d }t ||dS) NcWstjdS)N)nprandom default_rng)r*rrrr"sz2TestReadHtml.test_to_html_compat..F)Z data_gen_fZ c_idx_namesZ r_idx_namesz{:.3f}classZ dataframer)attrsrC) r'ZmakeCustomDataframer%formatZastypefloatto_htmlrrr()rIdfoutresrrrtest_to_html_compats z TestReadHtml.test_to_html_compatc sttdtjdgddtgdddtdtjdgddtgd ddgd gd gd gd d}|dkrOttjgd tjd}ttjddtgtjd}nt dt gd }t gd }|j dd}t d||jt||dd}Wdn1swYttdtjdgddtgdddtdtjdgddtgd ddtddtgddtgd dd||d|dkrddlddlmtfddjDt|dS)NrSZInt64)dtype)rarTrS?@ZFloat64)rcg@rd)TFN)TFTabc)rfrgN)rfrgrhdefghpythonrfrgpyarrowFindexzmode.string_storage dtype_backendrTboolean)ArrowExtensionArraycs$i|]}|j|ddqS)T)Z from_pandas)array).0colruexpectedparr sz3TestReadHtml.test_dtype_backend..)r r rUnanrrvZobject_r r7r8rr\pdZoption_contextrrroZ pandas.arraysrucolumnsr'r()rIZstring_storagersr]Z string_arrayZstring_array_nar^resultrryrtest_dtype_backendsR     zTestReadHtml.test_dtype_backendcCjt|dd }|j|d|j|jdd}|j|jdd}Wdn1s)wYt||dS)Nutf-8encodingcontentFirst Federal Bank of Floridar1 Metcalf Bankopen serve_contentreadrr?r-)rI httpserverrPrkdf1df2rrrtest_banklist_urls zTestReadHtml.test_banklist_urlcCr)Nrrr .*Water.*r1Unitr)rIrrMrkrrrrr test_spam_urls zTestReadHtml.test_spam_urlcCs6|j|dddid}|j|dddid}t||dS)Nz .*Florida.*idtabler2rYrrr-)rIrPrrrrr test_banklistszTestReadHtml.test_banklistcCsV|j|dd}|j|dd}t|||djddksJ|djddks)JdS)Nrr1rrrr ProximatesZNutrient)rr-ilocrrIrMrrrrr test_spams  zTestReadHtml.test_spamcCs&||}|D] }t|tsJqdSrrrr )rIrMdfsr]rrrtest_spam_no_matchs zTestReadHtml.test_spam_no_matchcCs.|j|ddid}|D] }t|tsJq dS)Nrr)rYr)rIrPrr]rrrtest_banklist_no_matchsz#TestReadHtml.test_banklist_no_matchcCs4|j|dddd}|jddksJ|jrJdS)NrrTr2headerrr)rrr))rIrMr]rrrtest_spam_headerszTestReadHtml.test_spam_headercC.|j|ddd}|j|ddd}t||dSNrrar2skiprowsrrrrrrtest_skiprows_intzTestReadHtml.test_skiprows_intcC6|j|dtdd}|j|dtdd}t||dSNrrTrr)rranger-rrrrtest_skiprows_range$z TestReadHtml.test_skiprows_rangecCs6|j|dddgd}|j|dddgd}t||dSNrrarTrrrrrrrtest_skiprows_list*rzTestReadHtml.test_skiprows_listcCs6|j|dddhd}|j|dddhd}t||dSrrrrrrtest_skiprows_set0rzTestReadHtml.test_skiprows_setcCrrrrrrrtest_skiprows_slice6rz TestReadHtml.test_skiprows_slicecCrrrslicer-rrrrtest_skiprows_slice_short<rz&TestReadHtml.test_skiprows_slice_shortcCs<|j|dtddd}|j|dtdddd}t||dS) NrrTrrrRrarrrrrtest_skiprows_slice_longBsz%TestReadHtml.test_skiprows_slice_longcCs:|j|dtdd}|j|dtdd}t||dSr)rrUZaranger-rrrrtest_skiprows_ndarrayHsz"TestReadHtml.test_skiprows_ndarraycCsDtjtdd|j|dddWddS1swYdS)Nz%is not a valid type for skipping rowsr1rZasdfr)r7r: TypeErrorr)rIrMrrrtest_skiprows_invalidNs"z"TestReadHtml.test_skiprows_invalidcCrNrrr2rCrrrrrr test_indexRszTestReadHtml.test_indexcC2|j|dddd}|j|dddd}t||dSNrrar)r2rrCrrrrrrtest_header_and_index_no_typesWz+TestReadHtml.test_header_and_index_no_typescCrrrrrrr test_header_and_index_with_types\rz-TestReadHtml.test_header_and_index_with_typescCrrrrrrrtest_infer_typesaszTestReadHtml.test_infer_typescCst|dd}t|}Wdn1swYt|dd}t|}Wdn1s3wY|j|dd}|j|dd}t||dSNzUTF-8rrr1r)rrrrr-)rIrMrkdata1data2rrrrrtest_string_iogszTestReadHtml.test_string_iocCsft|dd }|}Wdn1swY|jt|dd}|jt|dd}t||dSr)rrrrr-)rIrMrkrrrrrr test_stringrs  zTestReadHtml.test_stringcCst|dd}|j|dd}Wdn1swYt|dd}|j|dd}Wdn1s5wYt||dSr)rrr-)rIrMrkrrrrrtest_file_like{szTestReadHtml.test_file_likecCsP|jdddtjtdd|jdddWddS1s!wYdS)Nz#urlopen error unknown url type: gitcoder1zgit://github.comr)rr7r:rrrIrrrrtest_bad_url_protocols"z"TestReadHtml.test_bad_url_protocolcCsV|jdddtjttfdd|j|jddWddS1s$wYdS)NzName or service not knownrrzHTTP Error 404: NOT FOUNDr1r)rr7r:rr>rr?rrrrtest_invalid_urls"zTestReadHtml.test_invalid_urlcCsN|}|jttj|dddid}t|tsJ|D] }t|ts$JqdS)NZFirstrrr)rrospathabspathrlistr rIrPr?rr]rrr test_file_urlszTestReadHtml.test_file_urlcCsL|}tjtdd|j|dddidWddS1swYdS)NzNo tables foundr1rrZ tasdfablerr7r:r>r)rIrPr?rrrtest_invalid_table_attrss  "z%TestReadHtml.test_invalid_table_attrscKs|j|fdddid|S)NMetcalfrrrr)rIrr+rrr _bank_dataszTestReadHtml._bank_datacC*|j|ddgdd}t|jtsJdS)Nrrarrrrr rIrPr]rrrtest_multiindex_headerz#TestReadHtml.test_multiindex_headercCr)NrrarC)rrrqr rrrrtest_multiindex_indexrz"TestReadHtml.test_multiindex_indexcCs@|j|ddgddgdd}t|jtsJt|jtsJdS)Nrra)rrC)rrrr rqrrrrtest_multiindex_header_indexsz)TestReadHtml.test_multiindex_header_indexcC,|j|ddgddd}t|jtsJdSNrra)rrrrrrr&test_multiindex_header_skiprows_tuplesz3TestReadHtml.test_multiindex_header_skiprows_tuplescCrrrrrrrtest_multiindex_header_skiprowsrz,TestReadHtml.test_multiindex_header_skiprowscCsB|j|ddgddgddd}t|jtsJt|jtsJdS)Nrra)rrCr)rrrqr rrrrr%test_multiindex_header_index_skiprowssz2TestReadHtml.test_multiindex_header_index_skiprowscCsZ|}|jttj|ttdddid}t|tsJ|D] }t|t s*Jq!dS)NZFloridarrr) rrrrrrecompilerrr rrrrtest_regex_idempotencysz#TestReadHtml.test_regex_idempotencycCsHd}tjt|d|j|dddWddS1swYdS)Nz\(you passed a negative value\)r1ZWaterrrrrIrMr,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

rrIrrr python_docsszTestReadHtml.python_docscCs0|j|d|j|jdd}t|dksJdS)NrPythonr1ra)rrr?r#)rIrrrrrrtest_multiple_matches s z"TestReadHtml.test_multiple_matchescCsB|j|d|j|jdd}dd|D}t|ddgksJdS)Nrrr1cSsg|] }|jdddqS)rrrR)r)rwr]rrr sz7TestReadHtml.test_python_docs_table..ZPythZWhat)rrr?sorted)rIrrrzzrrrtest_python_docs_tables z#TestReadHtml.test_python_docs_tablecCs&d}|t|}t|dksJdS)z@ Make sure that read_html ignores empty tables. a
A B
1 2
raN)rrr#)rIr3rrrrtest_empty_tablesszTestReadHtml.test_empty_tablescCs>|tdd}tddgddggddgd }t||dS) Na
A B
1 2
3 4
rrarTrSrRABrrrrr r'r(rIrrzrrrtest_multiple_tbody6sz TestReadHtml.test_multiple_tbodycCs4|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
rHeaderfirstrrqNrrrrrtest_header_and_one_columnVsz'TestReadHtml.test_header_and_one_columncCs8|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_trpsz"TestReadHtml.test_thead_without_trcCsd}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 rZrrr'r()rIZ data_templateZ expected1Z expected2rrZresult1Zresult2rrrtest_tfoot_reads   zTestReadHtml.test_tfoot_readcCs8|jtdddd}tddggdd}t||dS)Na 
S I
text 1944
rrtextr)SIrrrrrr&test_parse_header_of_non_string_columnsz3TestReadHtml.test_parse_header_of_non_string_columnc sddlmfdd}|j|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.htmlrrrr shaper%replaceapplyrr'r() rIrPrrr]Z ground_trutholdnewZdfnewZgtnewZ convertedZ date_colsrrrtest_banklist_headers     z!TestReadHtml.test_banklist_headercCspd}t|dd }|}Wdn1swY||vs"J|j|dddidd}||vs6JdS)Nz Gold Canyonrrrrrr)rrrZ to_string)rIrPgcrkZraw_textr]rrrtest_gold_canyons  zTestReadHtml.test_gold_canyoncCs<|jtdddd}|jtdddd}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
rra
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
)rrr'r()rIrzrrrrtest_different_number_of_cols s" "$!z*TestReadHtml.test_different_number_of_colscCs8|tdd}tgdggdd}t||dS)NaZ
A B C
a b c
rre)rrCrrrrrrtest_colspan_rowspan_1Ssz#TestReadHtml.test_colspan_rowspan_1cC<|jtdddd}tgdggdd}t||dS)Na
X Y Z W
A B C
rr)rrrZr#)XzX.1Yr&Wrrrrrr test_colspan_rowspan_copy_valueslsz-TestReadHtml.test_colspan_rowspan_copy_valuescCr%)Na(
A B C
D
rr)rrrrD)rrzB.1zB.2r#rrrrrrtest_colspan_rowspan_both_not_1sz,TestReadHtml.test_colspan_rowspan_both_not_1cCs<|jtdddd}tddggddgd}t||dS)Nz
A B
C
rrr#rrrrrrrrtest_rowspan_at_end_of_rows z'TestReadHtml.test_rowspan_at_end_of_rowcCsB|jtdddd}tddgddggddgd}t||dS)Nz
A B
rrrrrrrrrrtest_rowspan_only_rowss  z#TestReadHtml.test_rowspan_only_rowscCX|tdd}tddgddggddgddggd}tdd gg|d }t||dS) Nam
A B
a b
1 2
rrrrfrgraZlevelscodesrTrrrr r r'r(rIrrrzrrr+test_header_inferred_from_rows_with_only_ths$z8TestReadHtml.test_header_inferred_from_rows_with_only_thcCsltdtdddi}|}|jt|dgdd}t||d|jt|dgdd}t||ddS)Ndate1/1/2001 ZperiodsrarZ parse_datesrC)r rr\rrr'r()rIr]rzr_rrrtest_parse_dates_lists z"TestReadHtml.test_parse_dates_listcCsrttddd}t|dd|ddd}|jt|dd d gid d }td|i}t||d dS) Nr6r7r8cS t|Sr)strr5rrrrr"  z7TestReadHtml.test_parse_dates_combine..cSr;r)r<timerrrrr" r=)r5r>datetimerarTr9r) r rr r%rrr\r'r()rIZ raw_datesr]r_Znewdfrrrtest_parse_dates_combines   z%TestReadHtml.test_parse_dates_combinecCs|dddd}tj|sJt|dtj|s%Jt|d|j|ddd d }|jd ks6Jd |jd vs?J|djt dksKJt |j ddsVJdS)Nrrr3wikipedia_states.htmlz is not a filez is an empty fileArizonararr)< Unnamedrsq mifloat64)rrFHzPN$A) rrisfilereprgetsizerrrrbrUallcloselocrIrrrrrrtest_wikipedia_states_tablesz(TestReadHtml.test_wikipedia_states_tablecCsp|dddd}|j|dddd}|jdksJd |jd d vs#J|jjd ks+Jt|jd ds6JdS)Nrrr3rArBrr)rC rErrarT)ZAlaska)z Total area[2]rFrH)rrrZnlevelsrUrLrMrNrrr test_wikipedia_states_multiindexs z-TestReadHtml.test_wikipedia_states_multiindexcCsF|jtdddgd}tddggtddgd }t|d|dS) NaK
AB
ab
rrarrfrg)Unnamed: 0_level_0r)zUnnamed: 1_level_0rr)rrr r from_tuplesr'r(rrrr%test_parser_error_on_empty_header_row%s z2TestReadHtml.test_parser_error_on_empty_header_rowcCsP|jtdddd}tddidgd}|djtdks Jt||dS) Na
Header
1100#101
#)decimalrrgClg0@rrG)rrr rbrUr'r(rrrrtest_decimal_rows>szTestReadHtml.test_decimal_rowsargFcCsLtd}tjt|d|j||dWddS1swYdS)NzPassing a bool to header is invalid. Use header=None for no header or header=int or list-like of ints to specify the row(s) making up the column namesr1r)rescaper7r:rr)rIrMrXr,rrrtest_bool_header_arg[s "z!TestReadHtml.test_bool_header_argcCs:|jtddtidd}tdddgi}t||dS)Na
a
0.763
0.244
rfrrz0.763z0.244)rrr<r r'r(rrrrtest_convertersfszTestReadHtml.test_converterscCs:|jtddgdd}tddtjgi}t||dS)Na
a
0.763
0.244
gZd;?)Z na_valuesrrfg"~j?rrr rUr}r'r(rrrrtest_na_valuesszTestReadHtml.test_na_valuescCspd}tdddgi}|jt|ddd}t||tdtjtjgi}|jt|ddd}t||dS) Na
a
N/A
NA
rfzN/Ar F)Zkeep_default_narT)r rrr'r(rUr})rIZ html_data expected_dfhtml_dfrrrtest_keep_default_nas z!TestReadHtml.test_keep_default_nacCsB|tdd}tddgtjtjggddgd}t||dS)Nak
A B
a b
rrfrgrrrr\rrrrtest_preserve_empty_rowss z%TestReadHtml.test_preserve_empty_rowscCr/) NaU
AB
ab
12
rrrrfrgrar0rTrr2r3rrr,test_ignore_empty_rows_when_inferring_headers$z9TestReadHtml.test_ignore_empty_rows_when_inferring_headercCsNtgdd}gdgdg|_|jdd}|t|d}t||dS)N))ZHillaryDr+)ZBernieJr+)ZDonaldER)r)rRZAgeZParty)NamezUnnamed: 1_level_1zUnnamed: 2_level_1Frpr)r rr\rrr'r()rIr^r3r_rrrtest_multiple_header_rowss z&TestReadHtml.test_multiple_header_rowscCs@|dddd}|j|dd}t|tsJt|dtsJdS)Nrrr3rBrr)rrrr )rIrrDrrrrtest_works_on_valid_markupsz'TestReadHtml.test_works_on_valid_markupcCs&|dddd}|j|dddgddS) Nrrr3rOrrAr/r=r)rIrrPrrrtest_fallback_successsz"TestReadHtml.test_fallback_successcCs>tddd}ttjdd|d}|}d|vsJdS)Nz 2000-01-01r7r8rT)r7rRrp)rr rUrVrWZstandard_normalr\)rIrngr]rrrrtest_to_html_timestamps z#TestReadHtml.test_to_html_timestampcCstdddg}|}|jdd}|jdd}|jdd}|jdd}|jdd}d|vs0J||ks6J||ks
foo bar baz qux
foo
displayed_onlyrra)rrr'r(r#)rIrqZexp0Zexp1rrrrrtest_displayed_onlys  z TestReadHtml.test_displayed_onlyrqcCs>d}tt||dd}tddgddgd}t||dS) NaW
A B
1 2
4 5
rprrarRrTrrmr)rIrqZ html_tablerrzrrr&test_displayed_only_with_many_elements>sz3TestReadHtml.test_displayed_only_with_many_elementsz\ignore:You provided Unicode markup but also provided a value for from_encoding.*:UserWarningc Cstj|}tj|d}|d\}}z]t|d}|j||dd}Wdn1s2wYt|d}|jt ||dd}Wdn1sUwY|j||dd} t ||t || WdSt yt rd|vsd|vrtw)Nr_rb)rrCZ16Z32)rrbasenamesplitextsplitrrrpoprr'r( Exceptionrr7skip) rIr base_pathrootrtrZfobjZ from_stringZfrom_file_like from_filenamerrr test_encodeTs>      zTestReadHtml.test_encodecCs|jjddkrtdGdddt}|d}||s!Jtjtdd||WddS1s9wYdS) Nr6rAzNot applicable for lxmlc@seZdZddZdS)zFTestReadHtml.test_parse_failure_unseekable..UnseekableStringIOcSrNFrrrrrseekable{zOTestReadHtml.test_parse_failure_unseekable..UnseekableStringIO.seekableN)__name__ __module__ __qualname__rrrrrUnseekableStringIOzs rz?
spameggs
z#passed a non-rewindable file objectr1)rkeywordsgetr7r{rr:r>)rIrbadrrrtest_parse_failure_unseekablets  "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)rIrrrr__init__s zBTestReadHtml.test_parse_failure_rewinds..MockFile.__init__cSs|jrdn|j}d|_|S)NrT)rr)rIsizerrrrrsz>TestReadHtml.test_parse_failure_rewinds..MockFile.readcSs d|_dSr)r)rIoffsetrrrseeks 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)rIrZgoodrrrrtest_parse_failure_rewindss z'TestReadHtml.test_parse_failure_rewindscCsGdddtj}|dddd}||j|fd}||j|fd}|||s0|r9 |s0|s0d|jurF|jusIJJdS)NcseZdZfddZZS)z@TestReadHtml.test_importcheck_thread_safety..ErrorThreadc sFztWnty}z ||_WYd}~dSd}~wwd|_dSr)superrunrzerr)rIr __class__rrrs zDTestReadHtml.test_importcheck_thread_safety..ErrorThread.run)rrrr __classcell__rrrr ErrorThreadsrrrr3rB)targetr*) threadingThreadrstartis_aliver)rIrrrDZhelper_thread1Zhelper_thread2rrrtest_importcheck_thread_safetys $z+TestReadHtml.test_importcheck_thread_safetycCsB|dddd}t|}||d}||d}t||dS)Nrrr3r4r)rrr'r()rIrZfile_path_string file_pathrrrrrtest_parse_path_objects z#TestReadHtml.test_parse_path_objectcCs4|tdd}tdggdgd}t||dS)Nz
A
word1
word2
rz word1 word2rrrrrrrtest_parse_br_as_spaces z#TestReadHtml.test_parse_br_as_space)r$bodyrr c Csd}gdgdgdgdgdgdd}|d }|d }|d }|d kr4|d }|d}|d}n|dkr=|d }n|dkrF|d}n|dkrN|d}|jt||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_linksrr)rrr ZfillnarUr}r'r() rIrXZ gh_13141_dataZgh_13141_expectedZdata_expZfoot_expZhead_exprrzrrrtest_extract_linkss2    zTestReadHtml.test_extract_linkscCsDd}tjt|dt|ddWddS1swYdS)NzY`extract_links` must be one of {None, "header", "footer", "body", "all"}, got "incorrect"r1Z incorrectrrrrrrtest_extract_links_bad% "z#TestReadHtml.test_extract_links_badcCs6d}|jt|ddd}tdgg}t||dS)Nz
Google.com
r$rr)z Google.comzhttps://google.comrrIrrrzrrr test_extract_links_all_no_header-s z-TestReadHtml.test_extract_links_all_no_headercCsDd}tjt|dtdddWddS1swYdS)NzPdtype_backend numpy is invalid, only 'numpy_nullable' and 'pyarrow' are allowed.r1testnumpyrrrrHrrrtest_invalid_dtype_backend<rz'TestReadHtml.test_invalid_dtype_backendcCsBd}|t|d}tddgddggddgd }t||dS) Na
A B
A1 B1
A2 B2
rA1ZB1A2ZB2rrrrrrrrtest_style_tagDszTestReadHtml.test_style_tag)frrrrJr7fixturerMrPrQr`rmarknetworkZ single_cpurrZslowrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr rrr!r"r$r*r,r-r.r4r:r@rOrQrTrW parametrizerZr[r]r`rarbrhrirjrlror rrrsfilterwarningsrrrrrrrrrrrrrrrrFfs    7                  0    & /  H#         "  D rF)5collections.abcr functoolsrrrrrpathlibrrr urllib.errorrrrUr7Z pandas.compatrZpandas.util._test_decoratorsutilZ_test_decoratorstdZpandasr~r r r r r rrrrZpandas._testingZ_testingr'Zpandas.core.arraysrrZpandas.io.commonrrrr-r<r@rErrrZ skip_if_norFrrrrs@    ,