�
�'�Yc @ s� d Z d d d g Z d d l Z d d l m Z m Z yD d d l m Z m Z m Z m
Z
m Z m Z m
Z
e e
f Z WnE e k
r� d d l m Z m Z m Z m
Z
m Z m Z e Z n Xd d d � Z d d d
� Z d d � Z d � Z e j d
e j � j Z d d d � � YZ d � Z d � Z y d d l m Z Wn! e k
rjd d l m Z n Xe j d � j Z y e! Wn e" k
r�e# Z! n Xd � Z$ d S( s5 External interface to the BeautifulSoup HTML parser.
t
fromstringt parset convert_treei����N( t etreet html( t
BeautifulSoupt Tagt Commentt ProcessingInstructiont NavigableStringt Declarationt Doctype( R R R R R R
c K s t | | | | � S( s� Parse a string of HTML data into an Element tree using the
BeautifulSoup parser.
Returns the root ``<html>`` 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.
( t _parse( t datat
beautifulsoupt makeelementt bsargs( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR s c K s@ t | d � s t | � } n t | | | | � } t j | � S( sY 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.
t read( t hasattrt openR R t ElementTree( t fileR R R t root( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR $ s c C s= t | | � } | j � } x | D] } | j | � q" W| S( s 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.
( t
_convert_treet getchildrent remove( t beautiful_soup_treeR R t childrent child( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR 3 s
c K s� | d k r t } n t | d � r@ d | k r@ d | d <q@ n t | d � rk d | k rk d | d <qk n | | | � } t | | � } t | � d k r� | d j d k r� | d Sd | _ | S( Nt
HTML_ENTITIESt convertEntitiesR t DEFAULT_BUILDER_FEATURESt featuress html.parseri i ( t NoneR R R t lent tag( t sourceR R R t treeR ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR E s % s` (?:\s|[<!])*DOCTYPE\s*HTML(?:\s+PUBLIC)?(?:\s+(\'[^\']*\'|"[^"]*"))?(?:\s+(\'[^\']*\'|"[^"]*"))?t
_PseudoTagc B s e Z d � Z d � Z RS( c C s d | _ g | _ | | _ d S( NR ( t namet attrst contents( t selfR) ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt __init__` s c C s
| j j � S( N( R) t __iter__( R* ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR, e s ( t __name__t
__module__R+ R, ( ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR& ^ s c C s� | d k r t j j } n d } } d } } x� t | � D]� \ } } t | t � r� | d k rl | } n | } | d k r� | j r� | j j � d k r� | } q� q< | d k r< t | t � r< | } q< q< W| d k r� g } } | j
}
n2 | j
| } | j
| | d !}
| j
| d } | d k rb|
j | � } |
| | j
|
| d | _
n t |
� } t
| � } | | � } | }
xB t | � D]4 } | | � } | d k r�|
j | � | }
q�q�W| }
x< | D]4 } | | � } | d k r�|
j | � | }
q�q�W| d k r�y | j � } Wn t k
rN| j } n Xt | � } | sdq�| j � \ } } | j � j } | o�| d d !| _ | o�| d d !| _ n | S( NR i i����( R! R t html_parserR t enumeratet
isinstanceR R' t lowert _DECLARATION_OR_DOCTYPER) t indexR& t _init_node_converterst reversedt addprevioust addnextt output_readyt AttributeErrort stringt _parse_doctype_declarationt groupst getroottreet docinfot public_idt
system_url( R R t first_element_idxt last_element_idxt html_roott declarationt it et pre_roott post_roott rootst convert_nodet res_roott prevt convertedt doctype_stringt matcht external_idt sys_uriR? ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR i sd
*
#
c s� i � g � � � f d � } � � f d � � d � � f d � � d � � d � � | t t � � � � � f d � � } | t � d � � } | t � d � � } | t � � f d � � } � S(
Nc s � � � f d � } | S( Nc s, x% � D] } | � | <� j | � q W| S( N( t append( t handlert t( t
converterst ordered_node_typest types( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt add� s
( ( RX RY ( RV RW ( RX s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt converter� s c s, x% � D] } t | | � r � | Sq Wd S( N( R1 R! ( t nodeRU ( RV RW ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt find_best_converter� s
c s_ y � t | � } Wn+ t k
rA � | � } � t | � <n X| d k rR d S| | | � S( N( t typet KeyErrorR! ( t bs_nodet parentRT ( RV R\ ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyRK � s
c S s� t | t � rf i } xd | j � D]= \ } } t | t � rO d j | � } n t | � | | <q" Wn t d � | D� � } | S( Nt c s s' | ] \ } } | t | � f Vq d S( N( t unescape( t .0t kt v( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pys <genexpr>� s ( R1 t dictt itemst listt joinRb ( t bs_attrst attribsRd Re ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt map_attrs� s c S sM t | � d k r+ | j p d | | _ n | d j p; d | | d _ d S( Ni t i����( R" t textt tail( R` Rn ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt append_text� s c s� | j } | d k rK | r' � | � n d } t j | | j d | �} n- | r] � | � n i } � | j d | �} xb | D]Z } y � t | � } Wn t k
r� n X| d k r | | | � q q � | | � q W| S( Nt attrib( R( R! R t
SubElementR' R] R^ ( R_ R` R( Rk t resR RT ( RK RV R Rl ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt convert_tag� s
c S s/ t j | � } | d k r+ | j | � n | S( N( R t HtmlCommentR! RS ( R_ R` Rs ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt convert_comment s c S sW | j d � r | d } n t j | j d d � � } | d k rS | j | � n | S( Nt ?i����Ra i ( t endswithR R t splitR! RS ( R_ R` Rs ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt
convert_pi s
c s& | d k r" � | t | � � n d S( N( R! Rb ( R_ R` ( Rp ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt convert_text s ( R! R R& R R R ( R RZ Rt Rv Rz R{ ( ( Rp RK RV R\ R Rl RW s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyR5 � s
' ( t name2codepoints &(\w+);c C s | s
d Sd � } t | | � S( NRm c S s= y t t | j d � � SWn t k
r8 | j d � SXd S( Ni i ( t unichrR| t groupR^ ( t m( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt unescape_entity5 s
( t handle_entities( R; R� ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyRb 1 s ( (% t __doc__t __all__t ret lxmlR R t bs4R R R R R R
R R3 t ImportErrorR! R R R R t compilet
IGNORECASERP R< R& R R5 t
html.entitiesR| t htmlentitydefst subR� R} t NameErrort chrRb ( ( ( s: /usr/lib64/python2.7/site-packages/lxml/html/soupparser.pyt <module> s: 4
.
U c