3
h>`C @ s d dddgZ ddlZddlZejdkrLddlZddlZddlZddlmZ nddl Z ddlmZ ddl
Z ddl Z ddl
T G d d
d
e jZejdkrG d d d ejjZnG d
d d e jZdd Zdd ZdddZdd ZdS ) escape_html
unescape_html html_to_text
html_document N )StringIO)*c @ s6 e Zd ZdddZdd Zdd Zd
d Zd d
ZdS )
TextWriterNP c C s( t jj| || d| _|| _| j d S )Nr ) Formatter
DumbWriter__init__ indent_level indent_width _set_indent)selffilemaxcolr r /usr/lib/python3.6/html_util.pyr 0 s zTextWriter.__init__c C s | j | j | _d| j | _d S )N )r r
indent_colindent)r r r r r 6 s zTextWriter._set_indentc C s || _ | j d S )N)r r )r Zmarginlevelr r r
new_margin: s zTextWriter.new_marginc C sF |d }t || jkr"| j| n | jt | }| jd| | d S )Nr )lenr Zsend_literal_data)r dataoffsetr r r send_label_data> s
zTextWriter.send_label_datac C s |sd S | j p|d tjk}| j}| j}| jj}| j}|dkrN|| j | j}xb|j D ]V}|r|t
| |kr|d| j | j}n|d |d }|| |t
| }d}qXW || _|d tjk| _ d S )Nr
r ) atbreakstringZ
whitespacecolr r writer r splitr )r r r# r% r r& Zwordr r r send_flowing_dataF s,
z TextWriter.send_flowing_data)Nr
r )__name__
__module__ __qualname__r r r r r( r r r r r . s
r c s6 e Zd Zd
fdd Zdd Zdd Zdd Z ZS ) HTMLParserAnchorFc s t t| j || _d | _d S )N)superr, r formatter anchor_href)r r. strictZconvert_charrefs) __class__r r r b s zHTMLParserAnchor.__init__c C s, |dkr(x |D ]\}}|dkr|| _ qW d S )Nahref)r/ )r tagZattrskeyvaluer r r handle_starttagg s z HTMLParserAnchor.handle_starttagc C s4 |dkr0| j d kr*| jjjd| j d d | _ d S )Nr2 ())r/ r. writerr( )r r4 r r r
handle_endtagm s
z HTMLParserAnchor.handle_endtagc C s | j jj| d S )N)r. r: r( )r r r r r handle_datas s z HTMLParserAnchor.handle_data)FF)r) r* r+ r r7 r; r<
__classcell__r r )r1 r r, ` s r, c @ s& e Zd Zd ddZdd Zdd ZdS )
r, r c C s t jj| || d S )N)htmllib
HTMLParserr )r r. verboser r r r y s zHTMLParserAnchor.__init__c C s
|| _ d S )N)anchor)r r3 nametyper r r
anchor_bgn| s zHTMLParserAnchor.anchor_bgnc C s | j r | jd| j d | _ d S )Nz (%s) )rA r< )r r r r
anchor_end s zHTMLParserAnchor.anchor_endN)r )r) r* r+ r rD rE r r r r r, w s
c C s^ | d kr d S y@| j dd} | j dd} | j dd} | j dd} | j d d
} W n Y nX | S ) N&z&z>'z'"z")replace)sr r r r s c C sX | d kr d S d| kr| S | j dd} | j dd} | j dd} | j dd } | j d
d} | S ) NrF z<rG z>rH z'rI z"rJ z&)rK )rL r r r r s r
c C s| yDt }tjt||}t|}|j| |j |j }|j |S tk
rv } zt j t j
d| d S d }~X nX d S )Nz cannot convert html to text: %s) r r ZAbstractFormatterr r, Zfeedclosegetvalue ExceptionsyslogZLOG_ERR)htmlr bufferr. parsertexter r r r s
c G s d}d}|}x| D ]}t |tjr,||7 }qt |ttfrRx\|D ] }||7 }q@W qt|r| }t |ttfrx |D ] }||7 }qtW q||7 }q||7 }qW ||7 }|S )a& Wrap the body components in a HTML document structure with a valid header.
Accepts a variable number of arguments of of which canb be:
* string
* a sequences of strings (tuple or list).
* a callable object taking no parameters and returning a string or sequence of strings.
zl
z
)
isinstanceZsixZ string_typestuplelistcallable)Zbody_componentsheadtaildocZbody_componentitemresultr r r r s$
)r )r )r
)__all__rP sys version_inforQ Z html.parserZ
html.entitiesior r> r. r r$ typesr
r rS r? r, r r r r r r r r s.
1