U
Sh5 @ s^ d Z ddlmZ ddlZddlZddlmZmZ ddlm Z ddl
m Z m Z m
Z
mZ G dd deZG d d
d
ejZG d d d eZG d
d deZG dd deZG dd deZG dd deZG dd deZG dd deZG dd deZG dd d eZG d d d eZ G d d d eZ G d!d" d"Z G d#d$ d$e Z d,d'd'd(d d)d*d+Z dS )-a A simple but complete HTML to Abstract Syntax Tree (AST) parser.
The AST can also reproduce the HTML text.
Example::
>> text = ''
>> ast = tokenize_html(text)
>> list(ast.walk(include_self=True))
[Root(''), Tag('div', {'class': 'note'}), Tag('p'), Data('text')]
>> str(ast)
''
>> str(ast[0][0])
'text
'
Note: optional tags are not accounted for
(see https://html.spec.whatwg.org/multipage/syntax.html#optional-tags)
) annotationsN)abcdeque)
HTMLParser)AnyCallableIterableIteratorc @ s@ e Zd ZdZdddddZedddd Zddd
d Zd S )
Attributez'This class holds the tags's attributes.str)keyreturnc C s | |dS )z+If self doesn't have the key it returns ''. )get)selfr r Q/root/rtd-docs/venv/lib/python3.8/site-packages/myst_parser/parsers/parse_html.py __getitem__! s zAttribute.__getitem__z list[str]r
c C s | d S )z!Return 'class' attribute as list.class)splitr r r r classes% s zAttribute.classesc C s d dd | D S )z0Return a htmlized representation for attributes. c s s" | ]\}}| d | dV qdS )z=""Nr ).0r valuer r r , s z$Attribute.__str__..)joinitemsr r r r __str__* s zAttribute.__str__N)__name__
__module__ __qualname____doc__r propertyr r r r r r r
s
r
c @ s@ e Zd ZdZdAdddddd Zed
d d d
Zedd ddZdBdddddZdd dddZ dd ddd Z
dd d d Z dd d d!Z d"d d#d$Z
dd dd%d&Zd d d'd(Zdd d)d*ZdCd+dd,d-d.Zdd d/d0Zd1dd2d3d4ZdDdd"d5d6d7ZdEddd d8d9d:ZdFdd?d@ZdS )GElementz]An Element of the xml/html document.
All xml/html entities inherit from this class.
r Nr z dict | NoneNone)nameattrr
c C s$ || _ t|pi | _d| _g | _dS )zInitialise the element.N)r( r
attrs_parent _childrenr r( r) r r r __init__5 s zElement.__init__zElement | Noner c C s | j S )zReturn parent.)r+ r r r r parent< s zElement.parentz
list[Element]c C s | j dd S )zReturn copy of children.Nr, r r r r childrenA s zElement.childrenFbool)r1 deepcopyc C sn g }t |D ]V\}}t|ts"t|r.| }|jd kr@| |_n|j| krXtd| || q || _d S )Nz&different parent already set for item ) enumerate
isinstancer& AssertionErrorr3 r+ appendr, )r r1 r3 Z new_childreniitemr r r reset_childrenF s
zElement.reset_childrenint)indexr
c C s
| j | S Nr0 r r< r r r r S s zElement.__getitem__)r< r9 c C sD t |tst|jd k r0|j| kr0td|| |_| j||S Nz"different parent already set for: )r5 r& r6 r+ r, __setitem__r r< r9 r r r r@ V s
zElement.__setitem__)r< c C s | j |S r= )r, __delitem__r> r r r rB ] s zElement.__delitem__c C s
| j S r= )r, __len__r r r r rC ` s zElement.__len__zIterator[Element]c c s | j E d H d S r= r0 r r r r __iter__c s zElement.__iter__c C sD t |tst|jd k r0|j| kr0td|| |_| j||S r? )r5 r& r6 r+ r, insertrA r r r rE f s
zElement.insertc C s0 | | j| j}| D ]}| }|| q|S )z#Recursively copy and remove parent.) __class__r( r* r3 r7 )r _copychildZ _copy_childr r r r3 m s
zElement.deepcopyc C s6 | j j d| j}| jr*|d| j7 }|d7 }|S )N(z, ))rF r! r( r* r textr r r __repr__u s
zElement.__repr__0dict[str, Callable[[Element, dict], str]] | None
tag_overridesr
c K s t dS )zReturns a HTML string representation of the element.
:param tag_overrides: Provide a dictionary of render function
for specific tag names, to override the normal render format
N)NotImplementedErrorr rP kwargsr r r render| s zElement.renderc C s | S r= rT r r r r r s zElement.__str__r )r9 r
c C s || kS r= r )r r9 r r r __eq__ s zElement.__eq__) include_selfr
c c s, |r
| V | D ]}|V | E dH qdS )z Walk through the xml/html AST.N)walk)r rW rH r r r rX s
z Element.walk)inplacerecurser
c C sF | }|s| }|dd |jD |rB|D ]}|jddd q.|S )zdReturn copy with all `Data` tokens
that only contain whitespace / newlines removed.
c S s( g | ] }t |tr |j d ks|qS )r )r5 Datadatastrip)r er r r
s
z!Element.strip..T)rY rZ )r3 r: r1 r] )r rY rZ elementrH r r r r] s z
Element.stripTzstr | type[Element]zIterable[str] | None)
identifierr* r rW rZ r
c # s |r | n| }|r"t| g|}t r8 fddn
fdd}|dk rTt|n|}|D ]R}||r\|dk r||jjsq\|pi D ]\} }
|j| |
kr q\q|V q\dS )z:Find all elements that match name and specific attributes.c s
t | S r= )r5 cra r r z Element.find..c s
| j kS r= r( rb rd r r re rf N)
rX itertoolschaininspectisclasssetissubsetr* r r ) r ra r* r rW rZ iteratorZ test_funcrH r r r rd r find s
z Element.find)r N)F)N)F)FF)NNFT)r! r" r# r$ r. r% r/ r1 r: r r@ rB rC rD rE r3 rM rT r rV rX r] ro r r r r r& / s4
r& c @ s e Zd ZdZddddZdS )RootzThe root of the AST tree.r r c s d fdd| D S )z6Returns a string HTML representation of the structure.r c 3 s | ]}|j f V qd S r= rU r rH rS r r r s z Root.render..)r r rS r rr r rT s z Root.renderNr! r" r# r$ rT r r r r rp s rp c @ s" e Zd ZdZddddddZdS ) TagzKRepresent xml/html tags under the form: ... .NrN r rO c sh r | j kr | j | S d| j | jr0dnd | j dd fdd| D d| j d S )Nc 3 s" | ]}|j f d i V qdS )rP NrU rq rS rP r r r s z Tag.render..z`NrN r rO c K sD |d k r"| j |kr"|| j | |S d| j | jr4dnd | j dS )Nrv r r z/>r( r* rR r r r rT s z XTag.render)Nrt r r r r ry s ry c @ s e Zd ZdZddddZdS )VoidTagzJRepresent tags with no children, only start tag, like `
`r r c K s" d| j | jrdnd | j dS )Nrv r r rw rz rs r r r rT s zVoidTag.renderNrt r r r r r{ s r{ c s> e Zd Zdd fdd ZddddZd ddd Z ZS )
TerminalElementr r\ c s t d || _d S )Nr )superr. r\ r r\ rF r r r. s zTerminalElement.__init__r c C s6 | j }t|dkr"|d d d }| jj d|dS )N z...rI rJ )r\ lenrF r! rK r r r rM s zTerminalElement.__repr__c C s | | j}|S )zCopy and remove parent.)rF r\ )r rG r r r r3 s zTerminalElement.deepcopy)r! r" r# r. rM r3
__classcell__r r r r r| s r| c @ s e Zd ZdZddddZdS )r[ z8Represent data inside xml/html documents, like raw text.r r c K s | j S r= r} rs r r r rT
s z Data.renderNrt r r r r r[ s r[ c @ s e Zd ZdZddddZdS ) Declarationz.Represent declarations, like ``r r c K s d| j dS )Nzr} rs r r r rT s zComment.renderNrt r r r r r s r c @ s e Zd ZdZddddZdS )Piz`r r c K s d| j dS )Nz s.
AF