n1k d Z g dZddlZddlmZmZ ddlmZmZm Z m
Z
m Z m Z m
Z
e e
fZn # e$ r ddlmZmZm Z m
Z
m Z m Z e ZY nw xY wddZddZdd Zd
Z ej d ej j Z G d d
Zd Zd Z ddlm Z n# e$ r ddl m Z Y nw xY w ej d j Z e n
# e!$ r e"Z Y nw xY wd Z#dS )z5External interface to the BeautifulSoup HTML parser.
)
fromstringparse convert_tree N)etreehtml)
BeautifulSoupTagCommentProcessingInstructionNavigableString DeclarationDoctype)r r r
r r r
c t | ||fi |S )a Parse a string of HTML data into an Element tree using the
BeautifulSoup parser.
Returns the root ```` Element of the tree.
You can pass a different BeautifulSoup parser through the
`beautifulsoup` keyword, and a diffent Element factory function
through the `makeelement` keyword. By default, the standard
``BeautifulSoup`` class and the default factory of `lxml.html` are
used.
)_parse)data
beautifulsoup makeelementbsargss v/builddir/build/BUILD/imunify360-venv-2.6.2/opt/imunify360/venv/lib64/python3.11/site-packages/lxml/html/soupparser.pyr r s $
{ = =f = == c t | d st | } t | ||fi |}t j | S )aY Parse a file into an ElemenTree using the BeautifulSoup parser.
You can pass a different BeautifulSoup parser through the
`beautifulsoup` keyword, and a diffent Element factory function
through the `makeelement` keyword. By default, the standard
``BeautifulSoup`` class and the default factory of `lxml.html` are
used.
read)hasattropenr r ElementTree)filer r r roots r r r $ sK 4 Dzz $
{ = =f = =D T " ""r c t | | }| }|D ]}| | |S )a Convert a BeautifulSoup tree to a list of Element trees.
Returns a list instead of a single root Element to support
HTML-like soup with more than one root element.
You can pass a different Element factory through the `makeelement`
keyword.
)
_convert_tree getchildrenremove)beautiful_soup_treer r childrenchilds r r r 3 sN ,k : :D !!H E Or c |t }t |d r d|vrd|d< t |d r d|vrd|d< || fi |}t || }t | dk r|d j dk r|d S d|_ |S ) N
HTML_ENTITIESconvertEntitiesr DEFAULT_BUILDER_FEATURESfeaturesz html.parser r )r r r lentag)sourcer r r treer s r r r E s %
}o.. / F * *(.F$ %} 899 / V # #!.F: = * *6 * *D { + +D
4yyA~~$q'+//AwDH Kr z`(?:\s|[?OPQ?Q-QR'01A!1C1D1DE KK " " "2A2Y); ;eAaCDDk I u%% (55L |I&&H D
h
LOO Y ' ' ' D D
LOO LL # # # D 0 ( 5 5 7 7NN 0 0 0 ( /NNN 0 +>:: ;
#(<<>> K **,,4G + A AbD0AG !(!:WQrT]G Os F0 0GGc 0
i g fd} fd d
fd d
d |t t
fd } |t d } |t d } |t fd }S ) Nc fd}|S )Nc F D ] }| |< | | S r8 )append)handlert
convertersordered_node_typestypess r addz5_init_node_converters..converter..add s7
-
- '
1
"))!,,,,Nr r= )rl rm rj rk s ` r converterz(_init_node_converters..converter s.
r c F D ] }t | | r
| c S d S r8 )rB )noderi rj rk s r find_best_converterz2_init_node_converters..find_best_converter s> # % %A$""
%!!}$$$
%tr c t | }n-# t $ r | x}t | < Y nw xY w|d S || | S r8 )typeKeyError)bs_nodeparentrh rj rq s r r\ z+_init_node_converters..convert_node s| O g /GG O O O2E2Eg2N2N NGjg /// O ?4ww'''s 'AAc t | t rYi }| D ]A\ }}t |t rd | }t | ||< Bnt d | D }|S )N c 3 > K | ]\ }}|t | fV d S r8 unescape).0kvs r z;_init_node_converters..map_attrs.. s1 AA1Ax{{+AAAAAAr )rB dictitemslistjoinr{ )bs_attrsattribsr} r~ s r map_attrsz(_init_node_converters..map_attrs s h % % BG ((
)
)1 a&& $ A %a[[
)
AAAAAAAGr c t | dk r| j pd|z | _ d S | d j pd|z | d _ d S )Nr r? )r+ texttail)rv r s r append_textz*_init_node_converters..append_text sI v;;! !;,"4FKKK %bz 4" .convert_tag s
*/9ii&&&TG"67<HHHCC*/7ii&&&RG +gl7;;;C
%
%E
$T%[[1 &GE3'''
L $ $ $ $
s B
B
Bc \ t j | }|| | |S r8 )r HtmlCommentrg ru rv r s r convert_commentz._init_node_converters..convert_comment s/ w'' MM#
r c | d r
| d d } t j | dd }|| | |S )N?r? rx r* )endswithr r splitrg r s r
convert_piz)_init_node_converters..convert_pi s_ C # crclG)7==a+@+@A MM#
r c > | |t | d S r8 rz )ru rv r s r convert_textz+_init_node_converters..convert_text s( K 1 1 2 2 2tr r8 )r r0 r
r r ) r rn r r r r r r\ rj rq r rk s ` @@@@@@r rF rF sW J ( ( ( ( ( ( ( = = = YsJ , Yw Y$%% &% Y
r )name2codepointz&(\w+);c 0 | sdS d }t || S )Nr c t t | d S # t $ r | d cY S w xY w)Nr* r )unichrr grouprt )ms r unescape_entityz!unescape..unescape_entity5 sU .455 5 771:: s ,/ AA)handle_entities)rL r s r r{ r{ 1 s0 r
?F 3 33r )NNr8 )$__doc____all__relxmlr r bs4r r r
r r r
r rD ImportErrorr r r r compile
IGNORECASEra rM r0 r rF
html.entitiesr htmlentitydefssubr r NameErrorchrr{ r= r r r s 2
1
1 * +G 4 * * * * * > > > > # # # #
$ $ (RZ$ M ( ( ( ( ( ( ( (R R Rj^ ^ ^F.,,,,,,, . . .--------. "*Z((,
FF
FFF
4 4 4 4 4s3 + AAB
BB4B7 7C C