old-nlp/venv/lib/python3.7/site-packages/bs4/__pycache__/diagnose.cpython-37.pyc

B
 %<25>]6<00>@s
dZd
Z
ddlZddlmZ
ddlmZ
ddlZddlmZm	Z	
ddl
mZ
ddlZddl
Z
ddlZddlZddlZddlZddlZddlZdd	<09>Zd#dd<0C>
ZGd
d<0E>de<06>Zdd<10>ZdZdZd$dd<15>
Zd%dd<18>
Zd&dd<1B>
Zd'dd<1E>
Zd(d d!<21>
Zed"k<02>
reej<1F> <20><00>

dS))z=Diagnostic functions, mainly for use when doing tech support.<2E>MIT<49>N)
<01>StringIO)
<01>
HTMLParser)<02>
BeautifulSoup<75>__version__)
<01>builder_registryc
CsTtd
t
<00>

tdtj<00>

dddg}
x>|
D]6}x0tjD]}||jkr6Pq6W|
<01>|<02>

td|<00>

q*Wd|
kr<>|
<01>d<07>

y*dd	l	m
}
td
d<0B>tt
|j<0E><02>
<00>

Wn*tk
r<EFBFBD>
}
ztd<0C>

Wd
d
}~XYnXd|
k<06>
rydd
l}td|j
<00>

Wn,tk
<EFBFBD>
r
}
ztd<0F>

Wd
d
}~XYnXt|d<10><02>
r4|<00><12>}n<>|<00>d<11>
<01>
sL|<00>d<12>
<01>
rdtd|<00>

td<14>

d
Sy:tj<15>|<00>
<01>
r<>td|<00>

t|<00>
<01>}|<07><12>}Wd
QRXWntk
<EFBFBD>
r<>


YnXt<00>
x<>|
D]<5D>}td|<00>

d}	yt||d<18>}
d}	Wn8tk
<EFBFBD>r"
}
ztd|<00>

t<1B><1C>
Wd
d
}~XYnX|	<09>rBtd|<00>

t|
<EFBFBD><1D><00>

td<1C>

<00>
q<>Wd
S)z/Diagnostic suite for isolating common problems.z'Diagnostic running on Beautiful Soup %szPython version %szhtml.parser<65>html5lib<69>lxmlz;I noticed that %s is not installed. Installing it may help.zlxml-xmlr)
<01>etreezFound lxml version %s<>
.z.lxml is not installed or couldn't be imported.NzFound html5lib version %sz2html5lib is not installed or couldn't be imported.<2E>readzhttp:zhttps:z<"%s" looks like a URL. Beautiful Soup is not an HTTP client.zpYou need to use some other library to get the document behind the URL, and feed that document to Beautiful Soup.z7"%s" looks like a filename. Reading data from the file.z#Trying to parse your markup with %sF)
<01>featuresTz%s could not parse the markup.z#Here's what %s did with the markup:zP--------------------------------------------------------------------------------)<1E>printr<00>sys<79>versionrZbuildersr
<00>remove<76>appendr	r
<00>join<69>map<61>strZLXML_VERSION<4F>ImportErrorr<00>hasattrr<00>
startswith<EFBFBD>os<6F>path<74>exists<74>open<65>
ValueErrorr<00>	Exception<6F>	traceback<63>	print_excZprettify)<0B>dataZ
basic_parsers<72>name<6D>builderr
<00>
er<00>fp<66>parser<65>success<73>soup<75>r)<00>8/tmp/pip-install-_x9nvcel/beautifulsoup4/bs4/diagnose.py<70>diagnosesj



















































r+TcKsNd
dlm
}
x<|jt|<00>
f
d|
i
|<02><02>
D]\}}td||j|jf<00>

q(WdS)z<>Print out the lxml events that occur during parsing.

    This lets you see how lxml parses a document when no Beautiful
    Soup code is running.
    r)
r
<00>htmlz%s, %4s, %sN)r	r
<00>	iterparserr<00>tag<61>text)r!r,<00>kwargsr
<00>event<6E>elementr)r)r*<00>
lxml_traceYs
$
r3c@s`eZ
dZd
Zdd<03>Zdd<05>Zdd<07>Zdd	<09>Zd
d<0B>Zdd
<0A>Z	dd<0F>Z
dd<11>Zdd<13>Zdd<15>Z
dS)<17>AnnouncingParserz?Announces HTMLParser parse events, without doing anything else.cCst|
<01>

dS)
N)
r)<02>self<6C>
sr)r)r*<00>_pfs
zAnnouncingParser._pcCs|<00>d
|
<00>

dS)Nz%s START)
r7)r5r"<00>attrsr)r)r*<00>handle_starttagis
z AnnouncingParser.handle_starttagcCs|<00>d
|
<00>

dS)Nz%s END)
r7)r5r"r)r)r*<00>
handle_endtagls
zAnnouncingParser.handle_endtagcCs|<00>d
|
<00>

dS)Nz%s DATA)
r7)r5r!r)r)r*<00>handle_dataos
zAnnouncingParser.handle_datacCs|<00>d
|
<00>

dS)Nz
%s CHARREF)
r7)r5r"r)r)r*<00>handle_charrefrs
zAnnouncingParser.handle_charrefcCs|<00>d
|
<00>

dS)Nz%s ENTITYREF)
r7)r5r"r)r)r*<00>handle_entityrefus
z!AnnouncingParser.handle_entityrefcCs|<00>d
|
<00>

dS)Nz
%s COMMENT)
r7)r5r!r)r)r*<00>handle_commentxs
zAnnouncingParser.handle_commentcCs|<00>d
|
<00>

dS)Nz%s DECL)
r7)r5r!r)r)r*<00>handle_decl{s
zAnnouncingParser.handle_declcCs|<00>d
|
<00>

dS)Nz%s UNKNOWN-DECL)
r7)r5r!r)r)r*<00>unknown_decl~s
zAnnouncingParser.unknown_declcCs|<00>d
|
<00>

dS)Nz%s PI)
r7)r5r!r)r)r*<00>	handle_pi<70>s
zAnnouncingParser.handle_piN)<0E>__name__<5F>
__module__<EFBFBD>__qualname__<5F>__doc__r7r9r:r;r<r=r>r?r@rAr)r)r)r*r4cs
r4c
Cst<00>}
|
<01>
|<00>

d
S)z<>Print out the HTMLParser events that occur during parsing.

    This lets you see how HTMLParser parses a document when no
    Beautiful Soup code is running.
    N)r4<00>feed)r!r&r)r)r*<00>htmlparser_trace<63>s
rGZaeiouZbcdfghjklmnpqrstvwxyz<79>c
Cs>d
}
x4t|<00>
D](}|ddkr$t
}nt}|
t<03>|<03>
7}
qW|
S)z#Generate a random word-like string.<2E><00>r)<05>range<67>_consonants<74>_vowels<6C>random<6F>choice)<04>lengthr6<00>
i<>
tr)r)r*<00>rword<72>s




rS<00>c

Csd
<01>dd<03>t
|<00>
D<00>
<01>
S)z'Generate a random sentence-like string.<2E>
 c
ss|]}
tt
<01>dd
<01><02>
V
qdS)rT<00>	N)rSrN<00>randint)<02>.0rQr)r)r*<00>	<genexpr><3E>szrsentence.<locals>.<genexpr>)rrK)
rPr)r)r*<00>	rsentence<63>srZ<00><>c
Cs<>d
ddddddg}
g}x~t|<00>
D]r}t
<01>dd	<09>}|dkrRt
<01>|
<01>
}|<02>d
|<00>

q |dkrr|<02>tt
<01>dd<0C><02>
<01>

q |d
kr t
<01>|
<01>
}|<02>d|<00>

q Wdd<10>|<02>
dS)z+Randomly generate an invalid HTML document.<2E>
p<>div<69>spanrQ<00>
b<>script<70>tabler<00>z<%s><3E>
rTrJz</%s>z<html><3E>

z</html>)rKrNrWrOrrZr)<06>num_elementsZ	tag_names<65>elementsrQrOZtag_namer)r)r*<00>rdoc<6F>s











rg順
c

Cs(
td
t
<00>

t|<00>
}
tdt|
<01>
<00>

x<>dddgddgD]z}d}y"t<04><04>}t|
|<02>}t<04><04>}d}Wn6tk
r<EFBFBD>
}
ztd	|<00>

t<07><08>
Wd
d
}~XYnX|r6td|||f<00>

q6Wdd
l	m
}
t<04><04>}|<08>|
<01>

t<04><04>}td||<00>

dd
l}	|	<09>
<0A>}t<04><04>}|<02>|
<01>

t<04><04>}td||<00>

d
S)z.Very basic head-to-head performance benchmark.z1Comparative parser benchmark on Beautiful Soup %sz3Generated a large invalid HTML document (%d bytes).r	r,rzhtml.parserFTz%s could not parse the markup.Nz"BS4+%s parsed the markup in %.2fs.r)
r
z$Raw lxml parsed the markup in %.2fs.z(Raw html5lib parsed the markup in %.2fs.)rrrg<00>len<65>timerrrr r	r
ZHTMLrr<00>parse)
rer!r&r'<00>
ar(r_r$r
rr)r)r*<00>benchmark_parsers<72>s4
























rmr	cCsXt<00>
<01>}|j}t|<00>
}tt||
d
<01>}t<06>d|||<03>
t<08>	|<03>
}|<06>
d<03>

|<06>dd<05>
dS)N)<03>bs4r!r&zbs4.BeautifulSoup(data, parser)Z
cumulativez
_html5lib|bs4<73>2)<0C>tempfile<6C>NamedTemporaryFiler"rg<00>dictrn<00>cProfileZrunctx<74>pstatsZStatsZ
sort_statsZprint_stats)rer&Z
filehandle<EFBFBD>filenamer!<00>vars<72>statsr)r)r*<00>profile<6C>s





rx<00>__main__)
T)
rH)
rT)
r[)
rh)rhr	)!rE<00>__license__rs<00>ior<00>html.parserrrnrrZbs4.builderrrrtrNrprjrrr+r3r4rGrMrLrSrZrgrmrxrB<00>stdinrr)r)r)r*<00><module>
s8











C

!