o h@snddlmZddlmZddlmZmZddlZddlm Z ddl Z ddl Z ddl m Z ddlZddlZddlmZddlmmZddlZddlmZmZmZmZmZmZmZm Z m!Z!ddl"m#Z$dd l%m&Z&m'Z'dd l(m)Z)ej*gd d d dZ+ddZ,ddZ-ddZ.ddZ/ej*ej0de1de1dgdej0de1ddgd ddZ2GdddZ3dS))Iterator)partial)BytesIOStringION)Path)URLError)is_platform_windows) NA DataFrame MultiIndexSeries Timestamp date_rangeread_csv read_html to_datetime)ArrowStringArray StringArray)file_path_to_url)zchinese_utf-16.htmlzchinese_utf-32.htmlzchinese_utf-8.htmlzletz_latin1.html)paramscCs|ddd|jS)z6Parametrized fixture for HTML encoding test filenames.iodata html_encoding)param)requestdatapathrm/var/www/html/construction_image-detection-poc/venv/lib/python3.10/site-packages/pandas/tests/io/test_html.pyhtml_encoding_file(s rcOst|t|ksJdt|dt|d}ttdd||}|s(J|t||D]\}}tj||g|Ri||jrEJdq-dS)Nz*lists are not of equal size len(list1) == z, len(list2) == z$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_equal5s( r4cCsltd}td||ddtjtddt|ddd d dd WddS1s/wYdS) Nbs4html5lib __version__z4.2zPandas requires versionmatchrrhtml spam.htmlflavor)pytest importorskipsetattrraises ImportErrorr) monkeypatchrr5rrrtest_bs4_version_failsIs  "rDcCsZd}d}d|d}tjt|dtt|d|dWddS1s&wYdS)Nz google.comzinvalid flavorz\{z \} is not a valid set of flavorsr8googler9r=)r>rA ValueErrorrr)urlr=r0rrrtest_invalid_flavorRs  "rIcCsZtdtdtd|dddd}t|ddgd }t|ddgd }t||dS) Nr5lxmlr6rrr:valid_markup.htmlr) index_colr=)r>r?rr4)rfilenamedfs_lxmldfs_bs4rrrtest_same_ordering[s   rPr5r6)marksrJcCstt|jdS)Nr<)rrr)rrrrflavor_read_htmlfsrRc@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.r8a
A B
1 2
3 4
)r)assert_produces_warning FutureWarning)selfrRr0rrrtest_literal_html_deprecationqs"z*TestReadHtml.test_literal_html_deprecationcC|ddddS)Nrrr:r;rrVrrrr spam_datazTestReadHtml.spam_datacCrX)Nrrr: banklist.htmlrrYrrr banklist_datar[zTestReadHtml.banklist_datacCshttjddtjtdtdddj  t }| }|t |ddid d d }t||dS) N)abcdtypecolumnsz{:.3f}class dataframer)attrsrL)r nprandom default_rngpdIndexlistobjectr'formatastypefloatto_htmlrr)r*)rVrRdfoutresrrrtest_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)Nr`Int64rb)rxr^r`?@Float64)rzg@r{)TFN)TFTabc)r~rN)r~rrdefghpythonr~rpyarrowr)ArrowExtensionArrayFindexzmode.string_storage dtype_backendTbooleancs$i|]}|j|ddqS)T) from_pandas)array).0colrexpectedparr sz3TestReadHtml.test_dtype_backend..)r r rinanrrobject_r r>r? pandas.arraysrrrsrloption_contextrrrer)r*) rVstring_storagerrRrt string_arraystring_array_naruresultrrrtest_dtype_backends\       zTestReadHtml.test_dtype_backendcCft|dd}|j|d||jdd}||jdd}Wdn1s'wYt||dS)Nutf-8encodingcontentFirst Federal Bank of Floridar8 Metcalf Bankopen serve_contentreadrHr4)rV httpserverr]rRrdf1df2rrrtest_banklist_urls zTestReadHtml.test_banklist_urlcCr)Nrrr .*Water.*r8Unitr)rVrrZrRrrrrrr test_spam_urls zTestReadHtml.test_spam_urlcCs2||dddid}||dddid}t||dS)Nz .*Florida.*idtabler9rhrr4)rVr]rRrrrrr test_banklists  zTestReadHtml.test_banklistcCsR||dd}||dd}t|||djddksJ|djddks'JdS)Nrr8rrrr ProximatesNutrient)r4ilocrerVrZrRrrrrr test_spam s   zTestReadHtml.test_spamcCs$||}|D] }t|tsJqdSrr )rVrZrRdfsrtrrrtest_spam_no_matchszTestReadHtml.test_spam_no_matchcCs,||ddid}|D] }t|tsJq dS)Nrr)rhr )rVr]rRrrtrrrtest_banklist_no_matchsz#TestReadHtml.test_banklist_no_matchcCs2||dddd}|jddksJ|jrJdS)Nrr^r9headerrr)rer+)rVrZrRrtrrrtest_spam_headerszTestReadHtml.test_spam_headercC*||ddd}||ddd}t||dSNrrxr9skiprowsrrrrrrtest_skiprows_int"zTestReadHtml.test_skiprows_intcC2||dtdd}||dtdd}t||dSNrr^rr)ranger4rrrrtest_skiprows_range(z TestReadHtml.test_skiprows_rangecCs2||dddgd}||dddgd}t||dSNrrxr^rrrrrrrtest_skiprows_list.rzTestReadHtml.test_skiprows_listcCs2||dddhd}||dddhd}t||dSrrrrrrtest_skiprows_set4rzTestReadHtml.test_skiprows_setcCrrrrrrrtest_skiprows_slice:rz TestReadHtml.test_skiprows_slicecCrrslicer4rrrrtest_skiprows_slice_short@rz&TestReadHtml.test_skiprows_slice_shortcCs8||dtddd}||dtdddd}t||dS) Nrr^rrr_rxrrrrrtest_skiprows_slice_longFsz%TestReadHtml.test_skiprows_slice_longcCs6||dtdd}||dtdd}t||dSr)riaranger4rrrrtest_skiprows_ndarrayLsz"TestReadHtml.test_skiprows_ndarraycCsBtjtdd||dddWddS1swYdS)Nz%is not a valid type for skipping rowsr8rasdfr)r>rA TypeError)rVrZrRrrrtest_skiprows_invalidRs"z"TestReadHtml.test_skiprows_invalidcCrNrrr9rLrrrrrr test_indexVszTestReadHtml.test_indexcC.||dddd}||dddd}t||dSNrrxr)r9rrLrrrrrrtest_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-8rrr8r)rrrr4)rVrZrRrdata1data2rrrrrtest_string_ioks  zTestReadHtml.test_string_iocCsbt|dd }|}Wdn1swY|t|dd}|t|dd}t||dSr)rrrr4)rVrZrRrrrrrrr test_stringvs  zTestReadHtml.test_stringcCs~t|dd}||dd}Wdn1swYt|dd}||dd}Wdn1s3wYt||dSr)rr4)rVrZrRrrrrrrtest_file_likeszTestReadHtml.test_file_likecCsN|jdddtjtdd|dddWddS1s wYdS)Nz#urlopen error unknown url type: gitcoder8zgit://github.comr)rr>rArrVrrRrrrtest_bad_url_protocols"z"TestReadHtml.test_bad_url_protocolcCsT|jdddtjttfdd||jddWddS1s#wYdS)NzName or service not knownrrzHTTP Error 404: NOT FOUNDr8r)rr>rArrGrHrrrrtest_invalid_urls"zTestReadHtml.test_invalid_urlcCsL|}|ttj|dddid}t|tsJ|D] }t|ts#JqdS)NFirstrrr)rospathabspathr!rnr rVr]rRrHrrtrrr test_file_urlszTestReadHtml.test_file_urlcCsJ|}tjtdd||dddidWddS1swYdS)NzNo tables foundr8rr tasdfablerr>rArG)rVr]rRrHrrrtest_invalid_table_attrss  "z%TestReadHtml.test_invalid_table_attrscC0||dddiddgdd}t|jtsJdS)NMetcalfrrrrx)r9rhrr!rer rVr]rRrtrrrtest_multiindex_header z#TestReadHtml.test_multiindex_headercCr)Nrrrrrx)r9rhrL)r!rr rrrrtest_multiindex_indexrz"TestReadHtml.test_multiindex_indexcCsF||dddiddgddgdd}t|jtsJt|jts!JdS)Nrrrrrx)r9rhrrL)r!rer rrrrrtest_multiindex_header_indexsz)TestReadHtml.test_multiindex_header_indexcC2||dddiddgddd}t|jtsJdSNrrrrrx)r9rhrrrrrrr&test_multiindex_header_skiprows_tuplesz3TestReadHtml.test_multiindex_header_skiprows_tuplescCrrrrrrrtest_multiindex_header_skiprowsrz,TestReadHtml.test_multiindex_header_skiprowscCsH||dddiddgddgddd}t|jtsJt|jts"JdS)Nrrrrrx)r9rhrrLr)r!rr rerrrr%test_multiindex_header_index_skiprowssz2TestReadHtml.test_multiindex_header_index_skiprowscCsX|}|ttj|ttdddid}t|tsJ|D] }t|ts)Jq dS)NFloridarrr) rrrrrecompiler!rnr rrrrtest_regex_idempotencysz#TestReadHtml.test_regex_idempotencycCsFd}tjt|d||dddWddS1swYdS)Nz\(you passed a negative value\)r8Waterrrr)rVrZrRr0rrrtest_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

rrVrrr python_docsszTestReadHtml.python_docscCs.|j|d||jdd}t|dksJdS)NrPythonr8rx)rrHr%)rVrrrRrrrrtest_multiple_matches(s z"TestReadHtml.test_multiple_matchescCs@|j|d||jdd}dd|D}t|ddgksJdS)Nrrr8cSsg|] }|jdddqS)rrr_)r)rrtrrr 4sz7TestReadHtml.test_python_docs_table..PythWhat)rrHsorted)rVrrrRrzzrrrtest_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
rxN)rr%)rVrRr:rrrrtest_empty_tables7s zTestReadHtml.test_empty_tablescCs<|tdd}tddgddggddgd }t||dS) Na
A B
1 2
3 4
rrxr^r`r_ABrrerr r)r*rVrRrrrrrtest_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
rHeaderfirstrrNrrrrrtest_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)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 rprr)r*) rVrR data_template expected1 expected2rrresult1result2rrrtest_tfoot_reads   zTestReadHtml.test_tfoot_readcCs6|tdddd}tddggdd}t||dS)Na 
S I
text 1944
rrtextr')SIrdrrrrr&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 BankrDrC) pandas.io.htmlr>rr shaper'replaceapplyrr)r*) rVr]rrRrArt 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)rVr]rRgcrraw_textrtrrrtest_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
rrLa
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*)rVrRrrrrrtest_different_number_of_cols's" "$!z*TestReadHtml.test_different_number_of_colscCs6|tdd}tgdggdd}t||dS)NaZ
A B C
a b c
rr})rrCrdrrrrrtest_colspan_rowspan_1osz#TestReadHtml.test_colspan_rowspan_1cC:|tdddd}tgdggdd}t||dS)Na
X Y Z W
A B C
rr8)rrrZrZ)XzX.1Yr]Wrrrrrr test_colspan_rowspan_copy_valuessz-TestReadHtml.test_colspan_rowspan_copy_valuescCr\)Na(
A B C
D
rr8)rrrrD)rrzB.1zB.2rZrrrrrrtest_colspan_rowspan_both_not_1sz,TestReadHtml.test_colspan_rowspan_both_not_1cCs:|tdddd}tddggddgd}t||dS)Nz
A B
C
rr8rZrrrrrrrrtest_rowspan_at_end_of_rows z'TestReadHtml.test_rowspan_at_end_of_rowcCs@|tdddd}tddgddggddgd}t||dS)Nz
A B
rr8rrrrrrrrtest_rowspan_only_rowss  z#TestReadHtml.test_rowspan_only_rowscCV|tdd}tddgddggddgddggd}tdd gg|d }t||dS) Nam
A B
a b
1 2
rrrr~rrxlevelscodesr^rrr r r)r*rVrRrrerrrr+test_header_inferred_from_rows_with_only_ths$z8TestReadHtml.test_header_inferred_from_rows_with_only_thcCshtdtdddi}|}|t|dgdd}t||d|t|dgdd}t||ddS)Ndate1/1/2001 periodsrxr parse_datesrL)r rrsrr)r*)rVrRrtrrvrrrtest_parse_dates_lists z"TestReadHtml.test_parse_dates_listcCspttddd}t|dd|ddd}|t|dd d gid d }td|i}t||d dS) NrnrorpcS t|Sr)strrmr@rrrr$% z7TestReadHtml.test_parse_dates_combine..cSrur)rvtimer@rrrr$&rw)rmrxdatetimerxr^rrr)r rr r'rrsr)r*)rVrR raw_datesrtrvnewdfrrrtest_parse_dates_combine!s   z%TestReadHtml.test_parse_dates_combinecCs|dddd}tj|sJt|dtj|s%Jt|d||ddd d }|jd ks5Jd |jd vs>J|djtdksJJt |j ddsUJdS)Nrrr:wikipedia_states.htmlz is not a filez is an empty fileArizonarxrr)< Unnamedrsq mifloat64)rrHzPN$A) rrisfilereprgetsizerIrercriallcloselocrVrrRrrrrrtest_wikipedia_states_table/sz(TestReadHtml.test_wikipedia_states_tablecCsn|dddd}||dddd}|jdksJd |jd d vs"J|jjd ks*Jt|jd ds5JdS)Nrrr:r}r~rr)r rrrxr^)Alaska)z Total area[2]rr)rIrenlevelsrirrrrrr test_wikipedia_states_multiindex9s z-TestReadHtml.test_wikipedia_states_multiindexcCsD|tdddgd}tddggtddgd }t|d|dS) NaK
AB
ab
rrxr8r~r)Unnamed: 0_level_0r)zUnnamed: 1_level_0rrd)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
#)decimalrr!gClg0@r#r)rr rcrir)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 namesr8r8)r escaper>rAr)rVrZrrRr0rrrtest_bool_header_argws "z!TestReadHtml.test_bool_header_argcCs8|tddtidd}tdddgi}t||dS)Na
a
0.763
0.244
r~rErz0.763z0.244)rrvr r)r*rrrrtest_convertersszTestReadHtml.test_converterscCs8|tddgdd}tddtjgi}t||dS)Na
a
0.763
0.244
gZd;?) na_valuesrr~g"~j?rr rirr)r*rrrrtest_na_valuesszTestReadHtml.test_na_valuescCsld}tdddgi}|t|ddd}t||tdtjtjgi}|t|ddd}t||dS) Na
a
N/A
NA
r~zN/Ar F)keep_default_narT)r rr)r*rir)rVrR html_data expected_dfhtml_dfrrrtest_keep_default_nas z!TestReadHtml.test_keep_default_nacCs@|tdd}tddgtjtjggddgd}t||dS)Nak
A B
a b
rr~rrrrrrrrrtest_preserve_empty_rowss z%TestReadHtml.test_preserve_empty_rowscCrf) NaU
AB
ab
12
rrrr~rrxrgr^rrjrkrrr,test_ignore_empty_rows_when_inferring_headers$z9TestReadHtml.test_ignore_empty_rows_when_inferring_headercCsLtgdd}gdgdg|_|jdd}|t|d}t||dS)N))HillaryDrb)BernieJrb)DonaldER)r)rAgeParty)NamezUnnamed: 1_level_1zUnnamed: 2_level_1Frr)r rersrr)r*)rVrRrr:rrrrtest_multiple_header_rowss z&TestReadHtml.test_multiple_header_rowscCs>|dddd}||dd}t|tsJt|dtsJdS)Nrrr:rKrrX)r!rnr )rVrrRrMrrrrtest_works_on_valid_markups z'TestReadHtml.test_works_on_valid_markupcCs$|dddd}||dddgddS) Nrrr:r\rrJr6rFr)rVrrRr]rrrtest_fallback_successsz"TestReadHtml.test_fallback_successcCs>tddd}ttjdd|d}|}d|vsJdS)Nz 2000-01-01rorpr^)ror_r)rr rirjrkstandard_normalrs)rVrngrtrrrrtest_to_html_timestamps z#TestReadHtml.test_to_html_timestampcCstdddg}|}|jdd}|jdd}|jdd}|jdd}|jdd}d|vs0J||ks6J||ks
foo bar baz qux
foo
displayed_onlyrrx)rr)r*r%)rVrexp0exp1rRrrrrrtest_displayed_only5s  z TestReadHtml.test_displayed_onlyrcCs>d}|t||dd}tddgddgd}t||dS) NaW
A B
1 2
4 5
rrrxr_r^rrr)rVrrR 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)rrL1632)rrbasenamesplitextsplitrrpoprr)r* Exceptionrr>skip) rVrrR base_pathrootrrfobj from_stringfrom_file_like from_filenamerrr test_encoders>      zTestReadHtml.test_encodecCsz|jddkr tdGdddt}|d}||sJtjtdd ||WddS1s6wYdS) Nr=rJzNot 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 objectr8)keywordsgetr>rrrArG)rVrRrbadrrrtest_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)rVrrrr__init__s zBTestReadHtml.test_parse_failure_rewinds..MockFile.__init__cSs|jrdn|j}d|_|S)Nr0T)rr)rVsizerrrrrsz>TestReadHtml.test_parse_failure_rewinds..MockFile.readcSs d|_dSr)r)rVoffsetrrrseeks 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)rVrRrgoodrrrrtest_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)rVr __class__rrrs zDTestReadHtml.test_importcheck_thread_safety..ErrorThread.run)rrrr __classcell__rrrr ErrorThreadsrrrr:rK)targetr.) threadingThreadstartis_aliver)rVrrRrrMhelper_thread1helper_thread2rrrtest_importcheck_thread_safetys $z+TestReadHtml.test_importcheck_thread_safetycCs>|dddd}t|}||d}||d}t||dS)Nrrr:r;r)rr)r*)rVrrRfile_path_string file_pathrrrrrtest_parse_path_objects   z#TestReadHtml.test_parse_path_objectcCs2|tdd}tdggdgd}t||dS)Nz
A
word1
word2
rz word1 word2rrrrrrrtest_parse_br_as_spaces z#TestReadHtml.test_parse_br_as_space)r&bodyrr1c Csd}gdgdgdgdgdgdd}|d }|d }|d }|d kr4|d }|d}|d}n|dkr=|d }n|dkrF|d}n|dkrN|d}|t||dd}t||g|d} | tj} t|| dS)Na
HTTP FTP Linkless
Wikipedia SURROUNDING Debian TEXT Linkless
Footer Multiple links: Only first captured.
)HTTPFTPLinkless))rN)rN)rz'https://en.wiktionary.org/wiki/linkless) WikipediaSURROUNDING Debian TEXTr))rzhttps://en.wikipedia.org/)rzftp://ftp.us.debian.org/)rN)Footer$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_extractr r rr&r r r rr1r extract_linksrrd)rr fillnarirr)r*) rVrrR 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"r8 incorrectrr>rArGr)rVrZr0rrrtest_extract_links_badC "z#TestReadHtml.test_extract_links_badcCs4d}|t|ddd}tdgg}t||dS)Nz
Google.com
r&rr)z Google.comzhttps://google.comrrVrRrrrrrr 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.r8testnumpyrr)rVr0rrrtest_invalid_dtype_backendZrz'TestReadHtml.test_invalid_dtype_backendcCs@d}|t|d}tddgddggddgd }t||dS) Na
A B
A1 B1
A2 B2
rA1B1A2B2rrrrrrrrtest_style_tagbszTestReadHtml.test_style_tag)drrrrWr>fixturerZr]rwrmarknetwork single_cpurrslowrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr rrrrrr r$r+r7r<rSrWrYr[rarcrdrerlrtr|rrrr parametrizerrrrrrrrrrrr rrfilterwarningsrrrrrrrrrrr$rrrrrSps  <                   0    & /  H#         "  D rS)4collections.abcr functoolsrrrrrpathlibrr r urllib.errorrrrir> pandas.compatrpandas.util._test_decoratorsutil_test_decoratorstdpandasrlr r r r r rrrrpandas._testing_testingr)pandas.core.arraysrrpandas.io.commonrr%rr4rDrIrPr skip_if_norRrSrrrrs@    ,