__init__.pyc
1 o 2 a��cC � @ s� d Z ddlmZ g d�ZddlZddlZddlZddlmZ z 3 ddl m 4 Z 5 m Z W n e y9 ddl m 6 Z 7 m Z Y nw ddlmZ d d 8 lmZ d d lmZ zdd lmZ W n e yc dd lmZ Y nw ze W n eys eZY nw ze W n ey� eefZY nw d d� ZdZ ej dde id�Z ej dde id�Z ej dde id�Z e� d�Z!e� d�Z"e� d�Z#e�$dej%�j&Z'e�$d�j&Z(ej dde id�Z)e�$d�Z*e�$d ej%�j+Z,d d � Z-d d � Z.d!d"� Z/G d#d$� d$e �Z0G d%d&� d&e1�Z2G d'd(� d(e1�Z3e3d)d*d+�Z4e3d,d*d+�Z5e3d-d.d+�Z6e3d/d.d+�Z7e3d0d*d+�Z8e3d1d.d+�Z9G d2d3� d3e2ej:�Z;G d4d5� d5e2ej<�Z=G d6d7� d7e2ej>�Z?G d8d9� d9e2ej@�ZAG d:d;� d;ejB�ZCe�$ed<�ej%�jDZEe�$d<�Fd=�ej%�jDZGd~d>d?�ZH dd@dA�ZI ddBdC�ZJd�dDdE�ZKd�dFdG�ZLdHdI� ZMdJdK� ZNG dLdM� dMe=�ZOeOeCjPdN< d�dOdP�ZQdQdR� ZRG dSdT� dTe 9 �ZSG dUdV� dVe1�ZTG dWdX� dXe1�ZUG dYdZ� dZeUe=�ZVeVeCjPd[< G d\d]� d]eUe=�ZWeWeCjPd^< G d_d`� d`e�ZXG dadb� dbeY�ZZG dcdd� ddeY�Z[G dedf� dfe�Z\G dgdh� dheUe=�Z]e]eCjPdi< G djdk� dke=�Z^e^eCjPdl< dmdn� Z_dodp� Z`e�$dq�jaZbe�$dq�Fdr��jaZc * d�dtdu�Zdeedj �ed_ d�dvdw�ZeG dxdy� dyejf�ZfG dzd{� d{ejg�Zhd|d}� Zief� Zjeh� ZkdS )�z.The ``lxml.html`` tool set for HTML handling. 10 � )�absolute_import)�document_fromstring�fragment_fromstring�fragments_fromstring� 11 fromstring�tostring�Element�defs�open_in_browser� submit_form�find_rel_links� 12 find_class�make_links_absolute�resolve_base_href� iterlinks� rewrite_links�parseN)�partial)�MutableMapping� 13 MutableSet� )�etree� )r )�SetMixin��urljoinc C sB | s| S t jd dkrt�dtj�j}nt�dtj�j}|d| �S )Nr � z^(\s*)u'z^(\s*)b'z\1')�sys� version_info�re�compile�M�sub)�sr" � r$ ��C:\Users\Jacks.GUTTSPC\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\site-packages\lxml\html\__init__.py�__fix_docstringK s 14 r& z http://www.w3.org/1999/xhtmlz9descendant-or-self::a[@rel]|descendant-or-self::x:a[@rel]�x�� 15 namespacesz7descendant-or-self::option|descendant-or-self::x:optionz3descendant-or-self::form|descendant-or-self::x:formztdescendant-or-self::*[@class and contains(concat(' ', normalize-space(@class), ' '), concat(' ', $class_name, ' '))]z descendant-or-self::*[@id=$id]zstring()z&url\((["][^"]*["]|['][^']*[']|[^)]*)\)z@import "(.*?)"z%//label[@for=$id]|//x:label[@for=$id]z[^ ]+z%[^;=]*;\s*(?:url\s*=\s*)?(?P<url>.*)$c C s\ | d d� dkr| dd � dks | d d� dkr*| dd � dkr*| dd� |d fS | |fS )Nr �"������'r$ )r# �posr$ r$ r% �_unquote_matchj s @r. c C s0 t | t�r t|dd�S t | t�rt|dd�S |S )z1Convert the result back into the input type. 16 zutf-8)�encoding�unicode)� 17 issubclass�bytesr r0 )�typ�resultr$ r$ r% �_transform_resultq s 18 19 20 r5 c C s@ t | t�r | d dkr | dtt�d � tkr | �d�d S | S )Nr �{r �}r+ )� 21 isinstance� 22 basestring�len�XHTML_NAMESPACE�split)�tagr$ r$ r% �_nons| s 23 $r> c s` e Zd ZdZdd� Zdd� Zdd� Z� fdd �Zd 24 d � Zd d � Z dd� Z 25 dd� Z dd� Z � Z S )�Classesa* Provides access to an element's class attribute as a set-like collection. 26 Usage:: 27 28 >>> el = fromstring('<p class="hidden large">Text</p>') 29 >>> classes = el.classes # or: classes = Classes(el.attrib) 30 >>> classes |= ['block', 'paragraph'] 31 >>> el.get('class') 32 'hidden large block paragraph' 33 >>> classes.toggle('hidden') 34 False 35 >>> el.get('class') 36 'large block paragraph' 37 >>> classes -= ('some', 'classes', 'block') 38 >>> el.get('class') 39 'large paragraph' 40 c C s || _ t|jdd�| _d S )N�class� )� _attributesr �get�_get_class_value)�self� 41 attributesr$ r$ r% �__init__� s zClasses.__init__c C sR |rt �d|�rtd| ��| �� �� }||v rdS |�|� d�|�| jd< dS )z[ 42 Add a class. 43 44 This has no effect if the class is already present. 45 �\s�Invalid class name: %rN� r@ )r �search� 46 ValueErrorrD r< �append�joinrB �rE �value�classesr$ r$ r% �add� s 47 z Classes.addc sh � rt �d� �rtd� ��� fdd�| �� �� D �}|r'd�|�| jd< dS d| jv r2| jd= dS dS )zn 48 Remove a class if it is currently present. 49 50 If the class is not present, do nothing. 51 rH rI c s g | ]}|� kr|�qS r$ r$ )�.0�name�rP r$ r% � 52 <listcomp>� s �z#Classes.discard.<locals>.<listcomp>rJ r@ N)r rK rL rD r< rN rB rO r$ rU r% �discard� s 53 �zClasses.discardc s0 |rt �d|�rtd| ��tt| ��|� dS )zw 54 Remove a class; it must currently be present. 55 56 If the class is not present, raise a KeyError. 57 rH rI N)r rK rL �superr? �remove�rE rP �� __class__r$ r% rY � s zClasses.removec C s | � � }||v o ||�� v S �N)rD r< )rE rT rQ r$ r$ r% � __contains__� s zClasses.__contains__c C � t | �� �� �S r] )�iterrD r< �rE r$ r$ r% �__iter__� � zClasses.__iter__c C r_ r] )r: rD r< ra r$ r$ r% �__len__� rc zClasses.__len__c C sL | � � �� }d}|D ] }||vr|�|� d}q 58 |r$d�|�| jd< dS dS )z. 59 Add all names from 'values'. 60 FTrJ r@ N)rD r< rM rN rB )rE �valuesrQ �extendedrP r$ r$ r% �update� s 61 ��zClasses.updatec C s� |rt �d|�rtd| ��| �� �� }z |�|� d}W n ty- |�|� d}Y nw |r:d�|�| jd< |S | jd= |S )z� 62 Add a class name if it isn't there yet, or remove it if it exists. 63 64 Returns true if the class was added (and is now enabled) and 65 false if it was removed (and is now disabled). 66 rH rI FTrJ r@ ) r rK rL rD r< rY rM rN rB )rE rP rQ �enabledr$ r$ r% �toggle� s 67 68 ��zClasses.toggle)�__name__� 69 __module__� __qualname__�__doc__rG rR rW rY r^ rb rd rg ri � __classcell__r$ r$ r[ r% r? � s 70 r? c s� e Zd Zd+� fdd� Zedd� �Zejdd� �Zedd� �Zed d 71 � �Zed d � �Z ed d� �Z 72 edd� �Z e jdd� �Z e j dd� �Z dd� Z dd� Zdd� Zdd� Zdd � Zd d � Zd,d d!�Z " d-d#d$�Zd+d%d&�Zd'd(� Z " d.d)d*�Z� ZS )/� HtmlMixinNc s t t| ��||� dS )z�set(self, key, value=None) 73 74 Sets an element attribute. If no value is provided, or if the value is None, 75 creates a 'boolean' attribute without value, e.g. "<form novalidate></form>" 76 for ``form.set('novalidate')``. 77 N)rX ro �set)rE �keyrP r[ r$ r% rp � s z HtmlMixin.setc C � 78 t | j�S )zB 79 A set-like wrapper around the 'class' attribute. 80 )r? �attribra r$ r$ r% rQ � � 81 zHtmlMixin.classesc C sH t |t�sJ �|�� }|r| �d|� d S | �d�d ur"| jd= d S d S )Nr@ )r8 r? rD rp rC rs )rE rQ rP r$ r$ r% rQ s �c C s | � � jjS )z� 82 Returns the base URL, given when the page was parsed. 83 84 Use with ``urlparse.urljoin(el.base_url, href)`` to get 85 absolute URLs. 86 )Z getroottree�docinfo�URLra r$ r$ r% �base_url 87 s zHtmlMixin.base_urlc C � t | �S )z0 88 Return a list of all the forms 89 )� _forms_xpathra r$ r$ r% �forms s zHtmlMixin.formsc C � | j ddtid�d S )zt 90 Return the <body> element. Can be called from a child element 91 to get the document's head. 92 z//body|//x:bodyr' r( r ��xpathr; ra r$ r$ r% �body � zHtmlMixin.bodyc C r{ )zu 93 Returns the <head> element. Can be called from a child 94 element to get the document's head. 95 z//head|//x:headr' r( r r| ra r$ r$ r% �head# r zHtmlMixin.headc C s. | � d�}|s dS t| |d�}|sdS |d S )zN 96 Get or set any <label> element associated with this element. 97 �idN�r� r )rC � _label_xpath)rE r� r4 r$ r$ r% �label+ s 98 zHtmlMixin.labelc C sD | � d�}|s td| ��t|j�dkrtd| ��|�d|� d S )Nr� z9You cannot set a label for an element (%r) that has no idr� z5You can only assign label to a label element (not %r)�for)rC � TypeErrorr> r= rp )rE r� r� r$ r$ r% r� 9 s 99 ����c C s | j }|d ur |jd= d S d S )Nr� )r� rs )rE r� r$ r$ r% r� F s �c C s^ | � � }|dus 100 J �| jr(| �� }|du r |jpd| j |_n |jp#d| j |_|�| � dS )z� 101 Removes this element from the tree, including its children and 102 text. The tail text is joined to the previous element or 103 parent. 104 NrA )� getparent�tail� getprevious�textrY )rE �parent�previousr$ r$ r% � drop_treeL s zHtmlMixin.drop_treec C s� | � � }|dus 105 J �| �� }| jr.t| jt�r.|du r%|jp d| j |_n |jp)d| j |_| jrZt| �rC| d }|jp=d| j |_n|du rQ|jpKd| j |_n |jpUd| j |_|�| �}| dd� |||d �< dS )a] 106 Remove the tag, but not its children or text. The children and text 107 are merged into the parent. 108 109 Example:: 110 111 >>> h = fragment_fromstring('<div>Hello <b>World!</b></div>') 112 >>> h.find('.//b').drop_tag() 113 >>> print(tostring(h, encoding='unicode')) 114 <div>Hello World!</div> 115 NrA r+ r ) r� r� r� r8 r= r9 r� r: �index)rE r� r� �lastr� r$ r$ r% �drop_tag\ s 116 zHtmlMixin.drop_tagc s � � � � � fdd�t| �D �S )z] 117 Find any links like ``<a rel="{rel}">...</a>``; returns a list of elements. 118 c s" g | ] }|� d ��� � kr|�qS ��rel�rC �lower�rS �elr� r$ r% rV � s �z,HtmlMixin.find_rel_links.<locals>.<listcomp>)r� �_rel_links_xpath)rE r� r$ r� r% r | s zHtmlMixin.find_rel_linksc C s t | |d�S )z> 119 Find any elements with the given class name. 120 )� 121 class_name)� _class_xpath)rE r� r$ r$ r% r � s zHtmlMixin.find_classc G s: z t | |d�d W S ty |r|d Y S t|��w )a� 122 Get the first element in a document with the given id. If none is 123 found, return the default argument if provided or raise KeyError 124 otherwise. 125 126 Note that there can be more than one element with the same id, 127 and this isn't uncommon in HTML documents found in the wild. 128 Browsers return only the first match, and this function does 129 the same. 130 r� r )� _id_xpath� 131 IndexError�KeyError)rE r� �defaultr$ r$ r% �get_element_by_id� s �zHtmlMixin.get_element_by_idc C rx )zT 132 Return the text content of the tag (and the text in any children). 133 )�_collect_string_contentra r$ r$ r% � text_content� s zHtmlMixin.text_content�htmlc C s ddl m} |||d�| �S )a 134 Run the CSS expression on this element and its children, 135 returning a list of the results. 136 137 Equivalent to lxml.cssselect.CSSSelect(expr, translator='html')(self) 138 -- note that pre-compiling the expression can provide a substantial 139 speedup. 140 r )� CSSSelector)� 141 translator)�lxml.cssselectr� )rE �exprr� r� r$ r$ r% � cssselect� s 142 zHtmlMixin.cssselectTc s� � du r| j � � du rtd��|r| �� |dkr � fdd�}n |dkr+� fdd�}n|du r6� fdd�}ntd | ��| �|� dS ) 143 a� 144 Make all links in the document absolute, given the 145 ``base_url`` for the document (the full URL where the document 146 came from), or if no ``base_url`` is given, then the ``.base_url`` 147 of the document. 148 149 If ``resolve_base_href`` is true, then any ``<base href>`` 150 tags in the document are used *and* removed from the document. 151 If it is false then any such tag is ignored. 152 153 If ``handle_failures`` is None (default), a failure to process 154 a URL will abort the processing. If set to 'ignore', errors 155 are ignored. If set to 'discard', failing URLs will be removed. 156 Nz3No base_url given, and the document has no base_url�ignorec s$ zt � | �W S ty | Y S w r] �r rL ��href�rw r$ r% � link_repl� s 157 �z0HtmlMixin.make_links_absolute.<locals>.link_replrW c s" zt � | �W S ty Y d S w r] r� r� r� r$ r% r� � s 158 �c s 159 t � | �S r] r r� r� r$ r% r� � � 160 z(unexpected value for handle_failures: %r)rw r� r rL r )rE rw r �handle_failuresr� r$ r� r% r � s$ ��z HtmlMixin.make_links_absolutec C sN d}| j ddtid�}|D ] }|�d�}|�� q |s dS | j|d|d� dS )a� 161 Find any ``<base href>`` tag in the document, and apply its 162 values to all links found in the document. Also remove the 163 tag once it has been applied. 164 165 If ``handle_failures`` is None (default), a failure to process 166 a URL will abort the processing. If set to 'ignore', errors 167 are ignored. If set to 'discard', failing URLs will be removed. 168 Nz //base[@href]|//x:base[@href]r' r( r� F)r r� )r} r; rC r� r )rE r� � base_hrefZbasetags�br$ r$ r% r � s 169 � 170 171 172 �zHtmlMixin.resolve_base_hrefc c s� � t j}| �tj�D �]7}|j}t|j�}|dkrpd}d|v r+|�d�}|d|dfV dD ]}||v rH|�|�}|durAt ||�}|||dfV q-d|v rot 173 � |�d��D ]}|� d�}|duret ||�}|d||� � fV qUn|D ]}||v r�|||| dfV qr|dkr�|�dd ��� } | d 174 kr�|�d d �} 175 t| 176 �}|r�|� d �n| 177 �� } | r�t| |r�|� d �n| 178 �| ��\} } |d | | fV nQ|d kr�|�d�p�d } | �� dkr�|d|�d�dfV n5|dk�r|j�rdd� t|j�D �dd� t|j�D � }|�r|jdd� |D ] \}} |d| |fV �qd|v �rBtt|d ��}|�rB|ddd� D ]}t|� d�|� d��\} }|d| |fV �q*q 179 dS )a 180 Yield (element, attribute, link, pos), where attribute may be None 181 (indicating the link is in the text). ``pos`` is the position 182 where the link occurs; often 0, but sometimes something else in 183 the case of links in stylesheets or style tags. 184 185 Note: <base href> is *not* taken into account in any way. The 186 link you get is exactly the link in the document. 187 188 Note: multiple links inside of a single text string or 189 attribute value are returned in reversed order. This makes it 190 possible to replace or delete them from the text string value 191 based on their reported text positions. Otherwise, a 192 modification at one text position can change the positions of 193 links reported later on. 194 �objectN�codebaser )�classid�data�archive�metaz 195 http-equivrA �refresh�content�url�param� valuetype�refrP �stylec S s, g | ]}t |�d �|�d ��ddd� �qS )r Nr+ )r. �group�start�rS �matchr$ r$ r% rV ; s ��z'HtmlMixin.iterlinks.<locals>.<listcomp>c S s g | ] }|� d �|�d �f�qS )r )r� r� r� r$ r$ r% rV ? s ��T)�reverser+ r )r � 196 link_attrsr` r r rs r> r= rC r � _archive_re�finditerr� r� r� �_parse_meta_refresh_url�stripr. �findr� �_iter_css_urls�_iter_css_imports�sort�list)rE r� r� �attribsr= r� rs rP r� � 197 http_equivr� r� r- r� �urlsr� r$ r$ r% r � s� � 198 199 200 201 � 202 203 �� �� ���� 204 ��zHtmlMixin.iterlinksc C s |dur | j ||d� n|r| �� | �� D ]h\}}}}||�� �}||kr'q|du r8|du r3d|_n|j|= q|du rT|jd|� | |j|t|� d� } | |_q|�|�} 205 |sft| 206 �t|�krf|} n| 207 d|� | | 208 |t|� d� } |�|| � qdS )a� 209 Rewrite all the links in the document. For each link 210 ``link_repl_func(link)`` will be called, and the return value 211 will replace the old link. 212 213 Note that links may not be absolute (unless you first called 214 ``make_links_absolute()``), and may be internal (e.g., 215 ``'#anchor'``). They can also be values like 216 ``'mailto:email'`` or ``'javascript:expr'``. 217 218 If you give ``base_href`` then all links passed to 219 ``link_repl_func()`` will take that into account. 220 221 If the ``link_repl_func`` returns None, the attribute or 222 tag text will be removed completely. 223 N)r rA ) r r r r� r� rs r: rC rp ) rE Zlink_repl_funcr r� r� rs �linkr- Znew_link�new�curr$ r$ r% r R s0 � ( 224 $�zHtmlMixin.rewrite_linksr] )r� )NTN)TN)rj rk rl rp �propertyrQ �setterrw rz r~ r� r� �deleterr� r� r r r� r� r� r r r r rn r$ r$ r[ r% ro � sF 225 226 227 228 229 230 231 232 233 234 235 � 236 -Z�ro c @ s& e Zd ZdZdefdd�Zdd� ZdS )� _MethodFunca5 237 An object that represents a method on an element as a function; 238 the function takes either an element or an HTML string. It 239 returns whatever the function normally returns, or if the function 240 works in-place (and so returns None) it returns a serialized form 241 of the resulting document. 242 Fc C s || _ || _t|| j �j| _d S r] )rT �copy�getattrrm )rE rT r� Z source_classr$ r$ r% rG � s z_MethodFunc.__init__c O s� t |�}t|t�r d|v rtd| j ��t|fi |��}nd|v r'|�d�}n| j}|r1t�|�}t || j�}||i |��}|d u rGt 243 ||�S |S )Nr� zQThe keyword 'copy' can only be used with element inputs to %s, not a string input) �typer8 r9 r� rT r �popr� �deepcopyr� r5 )rE �doc�args�kw� result_typeZ make_a_copy�methr4 r$ r$ r% �__call__� s" 244 � 245 246 z_MethodFunc.__call__N)rj rk rl rm ro rG r� r$ r$ r$ r% r� � s r� r F)r� r r Tr r r c @ � e Zd ZdS )� HtmlCommentN�rj rk rl r$ r$ r$ r% r� � � r� c @ r� )� HtmlElementNr� r$ r$ r$ r% r� � r� r� c @ r� )�HtmlProcessingInstructionNr� r$ r$ r$ r% r� � r� r� c @ r� )� 247 HtmlEntityNr� r$ r$ r$ r% r� � r� r� c @ s& e Zd ZdZi Zddd�Zdd� ZdS )�HtmlElementClassLookupav A lookup scheme for HTML Element classes. 248 249 To create a lookup instance with different Element classes, pass a tag 250 name mapping of Element classes in the ``classes`` keyword argument and/or 251 a tag name mapping of Mixin classes in the ``mixins`` keyword argument. 252 The special key '*' denotes a Mixin class that should be mixed into all 253 Element classes. 254 Nc 255 C s� t j�| � |d u r| j�� }|rWi }|D ]"\}}|dkr.|�� D ] }|�|g ��|� q!q|�|g ��|� q|�� D ]\}}|� |t 256 �}t ||g �} t |j | i �||< q<|| _d S )N�*)r �CustomElementClassLookuprG �_default_element_classesr� �keys� 257 setdefaultrM �itemsrC r� �tupler� rj �_element_classes) 258 rE rQ �mixinsZmixersrT rP �nZ mix_basesr� �basesr$ r$ r% rG � s 259 � 260 z HtmlElementClassLookup.__init__c C sB |dkr | j �|�� t�S |dkrtS |dkrtS |dkr tS d S )N�element�comment�PI�entity)r� rC r� r� r� r� r� )rE � node_type�document� namespacerT r$ r$ r% �lookup� s z HtmlElementClassLookup.lookup�NN)rj rk rl rm r� rG r� r$ r$ r$ r% r� � s 261 262 r� z^\s*<(?:html|!doctype)�asciic K sx |d u rt }tj| |fi |��}|d u rt�d��|r*|�d�d u r*|�dtd�� |r:|�d�d u r:|�td�� |S )NzDocument is emptyr� r r~ )� html_parserr r � ParserErrorr� �insertr rM )r� �parserZensure_head_bodyr� rP r$ r$ r% r � s �r c K s� |du rt }t| t�r t| �sd�d�| d�d� } nt| �s$d| } t| f||d�|��}t|j�dks8J �dd � |D �}t |�d 263 ksMJ d || f ��|d }g }|re|j 264 re|j 265 � � ret � d |j 266 ��|j 267 rs|j 268 � � rs|�|j 269 � |�|� |S )aR Parses several HTML elements, returning a list of elements. 270 271 The first item in the list may be a string. 272 If no_leading_text is true, then it will be an error if there is 273 leading text, and it will always be a list of only elements. 274 275 base_url will set the document's base_url attribute 276 (and the tree's docinfo.URL). 277 Nz <html><body>r� z</body></html>z <html><body>%s</body></html>�r rw r� c S s g | ] }t |j�d kr|�qS )r~ )r> r= �rS �er$ r$ r% rV s z(fragments_fromstring.<locals>.<listcomp>r ztoo many bodies: %r in %rr zThere is leading text: %r)r� r8 r2 �_looks_like_full_html_bytes�encode� _looks_like_full_html_unicoder r> r= r: r� r� r r rM �extend) r� �no_leading_textrw r r� r� �bodiesr~ �elementsr$ r$ r% r s0 278 �� � 279 r c K s� |du rt }t|�}t| f|| |d�|��}|r<t|t�s d}t|�}|r:t|d t�r5|d |_|d= |�|� |S |sCt� d��t 280 |�dkrXt� dd� d d 281 � |D �� ��|d }|j rl|j � � rlt� d |j ��d|_ |S ) a 282 283 Parses a single HTML element; it is an error if there is more than 284 one element, or if anything but whitespace precedes or follows the 285 element. 286 287 If ``create_parent`` is true (or is a tag name) then a parent node 288 will be created to encapsulate the HTML in a single element. In this 289 case, leading or trailing text is also allowed, as are multiple elements 290 as result of the parsing. 291 292 Passing a ``base_url`` will set the document's ``base_url`` attribute 293 (and the tree's docinfo.URL). 294 N)r r 295 rw �divr zNo elements foundr z Multiple elements found (%s)�, c S � g | ]}t |��qS r$ )� _element_namer r$ r$ r% rV Q � z'fragment_fromstring.<locals>.<listcomp>z Element followed by text: %r)r� �boolr r8 r9 r r� r r r r: rN r� r� ) r� Z create_parentrw r r� Zaccept_leading_textr �new_rootr� r$ r$ r% r ) sH ��� 296 297 298 299 ���r c K s� |du rt }t| t�rt| �}nt| �}t| f||d�|��}|r#|S |�d�}|s1|�dt �}|rn|d }t|�dkrm|dd� D ])}|j rct|�rZ|d j 300 pRd|j |d _ 301 n |j p^d|j |_ |� |� |� � qCnd}|�d �} | s~|�d 302 t �} | r�| d } 303 t| �dkr�| dd� D ] } | 304 � | � | � � q�|S |du r�|S t|�dkr�|j r�|j � � s�|d j 305 r�|d j 306 � � s�|d S t|�r�d |_|S d |_|S ) a 307 Parse the html, returning a single element/document. 308 309 This tries to minimally parse the chunk of text, without knowing if it 310 is a fragment or a document. 311 312 base_url will set the document's base_url attribute (and the tree's docinfo.URL) 313 Nr r~ z{%s}bodyr r r+ rA r� z{%s}headr �span)r� r8 r2 r r r �findallr; r: r� r� r r� r� �_contains_block_level_tagr= ) r� rw r r� Z is_full_htmlr� r r~ Z 314 other_body�headsr� Z 315 other_headr$ r$ r% r Z s\ 316 317 318 319 320 � 321 322 323 � ��r c K s$ |du rt }tj| |fd|i|��S )a= 324 Parse a filename, URL, or file-like object into an HTML document 325 tree. Note: this returns a tree, not an element. Use 326 ``parse(...).getroot()`` to get the document root. 327 328 You can override the base URL with the ``base_url`` keyword. This 329 is most useful when parsing from a file-like object. 330 Nrw )r� r r )Zfilename_or_urlr rw r� r$ r$ r% r � s r c C s, | � tj�D ] } t| j�tjv r dS qdS �NTF)r` r r r> r= r � 331 block_tags�r� r$ r$ r% r � s 332 �r c C s( t | tj�rdS t | t�rdS t| j�S )Nr� �string)r8 r � CommentBaser9 r> r= r r$ r$ r% r � s 333 334 335 r c @ s� e Zd ZdZedd� �Zedd� �Zejdd� �Zdd� Zd d 336 � Z ed d � �Z 337 e 338 jd d � �Z 339 e 340 j dd � �Z 341 edd� �Z e jdd� �Z dS )� FormElementz& 342 Represents a <form> element. 343 c C rx )z� 344 Returns an accessor for all the input elements in the form. 345 346 See `InputGetter` for more information about the object. 347 )� InputGetterra r$ r$ r% �inputs� s zFormElement.inputsc C rr )z� 348 Dictionary-like object that represents all the fields in this 349 form. You can set values in this dictionary to effect the 350 form. 351 )� 352 FieldsDictr ra r$ r$ r% �fields� s 353 zFormElement.fieldsc C sZ | j }|�� }|�� D ]\}}||v r|�|� |||< q |D ] }|d u r&q d ||< q d S r] )r! r� r� rY )rE rP r! Z prev_keysrq r$ r$ r% r! � s 354 355 356 �c C sb | � d�r 357 | � d�S | � d�rd| � d� S | jj}t|d��}|s*t|dt ��}t|�| ��S )NrT r� �#�formz{%s}form)rC r~ r` r� r; �strr� )rE Z iter_tagsrz r$ r$ r% �_name� s 358 359 360 zFormElement._namec C s� g }| j D ]i}|j}|rd|jv rqt|j�}|dkr$|�||jf� q|dkrH|j}|jr;|D ] }|�||f� q0q|durG|�||jf� q|dksRJ d| ��|jrY|j sYq|j 361 dv r_q|j}|durn|�||jf� q|S )z� 362 Return a list of tuples of the field values for the form. 363 This is suitable to be passed to ``urllib.urlencode()``. 364 �disabled�textarea�selectN�inputzUnexpected tag: %r)�submit�image�reset�file) r rT rs r> r= rM rP �multiple� checkable�checkedr� )rE �resultsr� rT r= rP �vr$ r$ r% � form_values� s: 365 366 �� 367 � 368 �zFormElement.form_valuesc C s* | j }| �d�}|r|durt||�S |S )z: 369 Get/set the form's ``action`` attribute. 370 �actionN)rw rC r )rE rw r4 r$ r$ r% r4 s 371 372 373 zFormElement.actionc C � | � d|� d S �Nr4 �rp rZ r$ r$ r% r4 � c C � | j }d|v r |d= d S d S r6 �rs �rE rs r$ r$ r% r4 # � 374 �c C � | � dd��� S )zt 375 Get/set the form's method. Always returns a capitalized 376 string, and defaults to ``'GET'`` 377 �method�GET)rC �upperra r$ r$ r% r> ) s zFormElement.methodc C s | � d|�� � d S )Nr> )rp r@ rZ r$ r$ r% r> 1 s N) rj rk rl rm r� r r! r� r% r3 r4 r� r> r$ r$ r$ r% r � s( 378 379 380 381 382 383 384 r r# c C sV | � � }|rt|d�r|�� }|�|� |du rt}| jr!| j}n| j}|| j||�S )a% 385 Helper function to submit a form. Returns a file-like object, as from 386 ``urllib.urlopen()``. This object also has a ``.geturl()`` function, 387 which shows the URL if there were any redirects. 388 389 You can use this like:: 390 391 form = doc.forms[0] 392 form.inputs['foo'].value = 'bar' # etc 393 response = form.submit() 394 doc = parse(response) 395 doc.make_links_absolute(response.geturl()) 396 397 To change the HTTP requester, pass a function as ``open_http`` keyword 398 argument that opens the URL for you. The function must have the following 399 signature:: 400 401 open_http(method, URL, values) 402 403 The action is one of 'GET' or 'POST', the URL is the target URL as a 404 string, and the values are a sequence of ``(name, value)`` tuples with the 405 form data. 406 r� N)r3 �hasattrr� r �open_http_urllibr4 rw r> )r# Z extra_values� open_httpre r� r$ r$ r% r 9 s 407 408 r c C s� |st d��z 409 ddlm}m} W n ty% ddlm} ddlm} Y nw | dkr@d|v r3|d7 }n|d7 }|||�7 }d }n||�}t|t�sN|� d �}|||�S ) 410 Nz cannot submit, no URL providedr )� urlencode�urlopen)rE )rD r? �?�&�ASCII) 411 rL �urllibrD rE � ImportError�urllib.request� urllib.parser8 r2 r )r> r� re rD rE r� r$ r$ r% rB _ s$ � 412 413 414 415 rB c @ sT e Zd Zdd� Zdd� Zdd� Zdd� Zd d 416 � Zd d � Zd d� Z dd� Z 417 dd� Z dS )r c C � 418 || _ d S r] �r )rE r r$ r$ r% rG x r� zFieldsDict.__init__c C s | j | jS r] �r rP �rE �itemr$ r$ r% � __getitem__z s zFieldsDict.__getitem__c C s || j | _d S r] rO )rE rQ rP r$ r$ r% � __setitem__| rc zFieldsDict.__setitem__c C s t d��)Nz'You cannot remove keys from ElementDict)r� rP r$ r$ r% � __delitem__~ s �zFieldsDict.__delitem__c C s 419 | j �� S r] )r r� ra r$ r$ r% r� � r� zFieldsDict.keysc C s 420 || j v S r] rN rP r$ r$ r% r^ � r� zFieldsDict.__contains__c C s t | j�� �S r] )r` r r� ra r$ r$ r% rb � s zFieldsDict.__iter__c C rr r] )r: r ra r$ r$ r% rd � r� zFieldsDict.__len__c C s d| j j| jj�� f S �Nz<%s for form %s>)r\ rj r r# r% ra r$ r$ r% �__repr__� s 421 �zFieldsDict.__repr__N) rj rk rl rG rR rS rT r� r^ rb rd rV r$ r$ r$ r% r v s r c @ sP e Zd ZdZdd� Zdd� Zdd� Zdd � Zd 422 d � Zd d � Z dd� Z 423 dd� Z dS )r a[ 424 An accessor that represents all the input fields in a form. 425 426 You can get fields by name from this, with 427 ``form.inputs['field_name']``. If there are a set of checkboxes 428 with the same name, they are returned as a list (a `CheckboxGroup` 429 which also allows value setting). Radio inputs are handled 430 similarly. Use ``.keys()`` and ``.items()`` to process all fields 431 in this way. 432 433 You can also iterate over this to get all input elements. This 434 won't return the same thing as if you get all the names, as 435 checkboxes and radio elements are returned individually. 436 c C rM r] )r# )rE r# r$ r$ r% rG � r� zInputGetter.__init__c C s d| j j| j�� f S rU )r\ rj r# r% ra r$ r$ r% rV � � �zInputGetter.__repr__c s� � fdd�| D �}|st d� ��|d �d�}|dkr+t|�dkr+t|�}� |_|S |dkr>t|�dkr>t|�}� |_|S |d S ) Nc s g | ] }|j � kr|�qS r$ �rT )rS �fieldrX r$ r% rV � s z+InputGetter.__getitem__.<locals>.<listcomp>z!No input element with the name %rr r� �radior �checkbox)r� rC r: � 437 RadioGrouprT � CheckboxGroup)rE rT r! Z 438 input_typer� r$ rX r% rR � s zInputGetter.__getitem__c C s | D ] 439 }|j |kr dS qdS r rX )rE rT rY r$ r$ r% r^ � s 440 441 �zInputGetter.__contains__c C s: g }dh}| D ]}|j }||vr|�|� |�|� q|S )zx 442 Returns all unique field names, in document order. 443 444 :return: A list of all unique field names. 445 N)rT rM rR )rE �names�seenr� rT r$ r$ r% r� � s 446 447 �zInputGetter.keysc C sB g }t � }| D ]}|j}||vr |�|� |�|| | f� q|S )z� 448 Returns all fields with their names, similar to dict.items(). 449 450 :return: A list of (name, field) tuples. 451 )rp rT rR rM )rE r� r_ r� rT r$ r$ r% r� � s 452 �zInputGetter.itemsc C s | j �ddd�S )Nr( r) r' )r# r` ra r$ r$ r% rb � rc zInputGetter.__iter__c C s t dd� | D ��S )Nc s s � | ]}d V qdS )r Nr$ )rS �_r$ r$ r% � <genexpr>� s � z&InputGetter.__len__.<locals>.<genexpr>)�sumra r$ r$ r% rd � s zInputGetter.__len__N) rj rk rl rm rG rV rR r^ r� r� rb rd r$ r$ r$ r% r � s r c @ s@ e Zd ZdZedd� �Zejdd� �Zejdd� �Zdd� ZdS ) � 453 InputMixinzE 454 Mix-in for all input elements (input, select, and textarea) 455 c C s 456 | � d�S )z1 457 Get/set the name of the element 458 rT �rC ra r$ r$ r% rT � rt zInputMixin.namec C r5 �NrT r7 rZ r$ r$ r% rT � r8 c C r9 re r: r; r$ r$ r% rT � r< c C s8 t | dd �}|r d| }nd}d| jjt| �| j|f S )Nr� z type=%rrA z<%s %x name=%r%s>)r� r\ rj r� rT )rE � type_namer$ r$ r% rV � s 459 �zInputMixin.__repr__N) rj rk rl rm r� rT r� r� rV r$ r$ r$ r% rc � s 460 461 462 rc c @ �8 e Zd ZdZedd� �Zejdd� �Zejdd� �ZdS )�TextareaElementzp 463 ``<textarea>`` element. You can get the name with ``.name`` and 464 get/set the value with ``.value`` 465 c C sF | j pd}| j�dt �rd}nd}| D ] }|tj||dd�7 }q|S )zK 466 Get/set the value (which is the contents of this element) 467 rA �{%s}�xmlr� r0 �r> r/ )r� r= � 468 startswithr; r r )rE r� Zserialisation_methodr� r$ r$ r% rP s 469 470 �zTextareaElement.valuec C s | d d �= || _ d S r] �r� rZ r$ r$ r% rP s 471 472 c C s d| _ | d d �= d S )NrA rm ra r$ r$ r% rP " s N)rj rk rl rm r� rP r� r� r$ r$ r$ r% rh s 473 474 rh r' c @ s^ e Zd ZdZedd� �Zejdd� �Zejdd� �Zedd� �Zedd � �Z e jd 475 d � �Z d S ) � SelectElementa� 476 ``<select>`` element. You can get the name with ``.name``. 477 478 ``.value`` will be the value of the selected option, unless this 479 is a multi-select element (``<select multiple>``), in which case 480 it will be a set-like object. In either case ``.value_options`` 481 gives the possible values. 482 483 The boolean attribute ``.multiple`` shows if this is a 484 multi-select. 485 c C s� | j rt| �S t| �}z tdd� t|�D ��}W n ty8 z tdd� |D ��}W n ty5 Y Y dS w Y nw |�d�}|du rI|jpFd�� }|S )z� 486 Get/set the value of this select (the selected option). 487 488 If this is a multi-select, this is a set-like object that 489 represents all the selected options. 490 c s s" � | ] }|� d �dur|V qdS )�selectedNrd r� r$ r$ r% ra D � � z&SelectElement.value.<locals>.<genexpr>c s s" � | ] }|� d �du r|V qdS )r& Nrd r� r$ r$ r% ra G rp NrP rA ) r. �MultipleSelectOptions�_options_xpath�next�reversed� StopIterationrC r� r� )rE �optionsZselected_optionrP r$ r$ r% rP 7 s" ��� 491 zSelectElement.valuec C s� | j rt|t�r td��| j}|�� |�|� d S d }|d urEt| �D ]}|�d�}|d u r6|j p3d� 492 � }||kr>|} nq$t d| ��t| �D ] }d|j v rT|j d= qI|d ura|� dd� d S d S )NzYou must pass in a sequencerP rA z'There is no option with the value of %rro )r. r8 r9 r� rP �clearrg rr rC r� r� rL rs rp )rE rP re �checked_optionr� � opt_valuer$ r$ r% rP O s6 493 494 495 �� 496 ��c C s | j r 497 | j�� d S d | _d S r] )r. rP rw ra r$ r$ r% rP j s 498 c C s@ g }t | �D ]}|�d�}|du r|jpd�� }|�|� q|S )z� 499 All the possible values this select can have (the ``value`` 500 attribute of all the ``<option>`` elements. 501 rP NrA )rr rC r� r� rM )rE rv r� rP r$ r$ r% � value_optionsr s 502 zSelectElement.value_optionsc C s 503 d| j v S )zW 504 Boolean attribute: is there a ``multiple`` attribute on this element. 505 r. r: ra r$ r$ r% r. � rt zSelectElement.multiplec C s. |r 506 | � dd� d S d| jv r| jd= d S d S )Nr. rA )rp rs rZ r$ r$ r% r. � s 507 508 �N) 509 rj rk rl rm r� rP r� r� rz r. r$ r$ r$ r% rn + s 510 511 512 513 514 rn r( c @ sD e Zd ZdZdd� Zedd� �Zdd� Zdd � Zd 515 d � Z d d � Z 516 dS )rq z� 517 Represents all the selected options in a ``<select multiple>`` element. 518 519 You can add to this set-like option to select an option, or remove 520 to unselect the option. 521 c C rM r] )r( )rE r( r$ r$ r% rG � r� z MultipleSelectOptions.__init__c C s t t| j��S )z< 522 Iterator of all the ``<option>`` elements. 523 )r` rr r( ra r$ r$ r% rv � � z MultipleSelectOptions.optionsc c sB � | j D ]}d|jv r |�d�}|d u r|jpd�� }|V qd S )Nro rP rA )rv rs rC r� r� )rE �optionry r$ r$ r% rb � s � 524 525 526 ��z MultipleSelectOptions.__iter__c C sR | j D ] }|�d�}|d u r|jpd�� }||kr"|�dd� d S qtd| ��)NrP rA ro z$There is no option with the value %r)rv rC r� r� rp rL �rE rQ r| ry r$ r$ r% rR � s 527 528 ��zMultipleSelectOptions.addc C sd | j D ](}|�d�}|d u r|jpd�� }||kr+d|jv r%|jd= d S td| ��qtd| ��)NrP rA ro z'The option %r is not currently selectedz%There is not option with the value %r)rv rC r� r� rs rL r} r$ r$ r% rY � s 529 530 531 ����z MultipleSelectOptions.removec C �& d| j jd�dd� | D ��| jjf S )Nz <%s {%s} for select name=%r>r c S r r$ ��repr�rS r2 r$ r$ r% rV � r z2MultipleSelectOptions.__repr__.<locals>.<listcomp>)r\ rj rN r( rT ra r$ r$ r% rV � � 532 �z MultipleSelectOptions.__repr__N) rj rk rl rm rG r� rv rb rR rY rV r$ r$ r$ r% rq � s 533 rq c @ �L e Zd ZdZedd� �Zejdd� �Zejdd� �Zedd� �Zdd � Z d 534 S ) r\ a 535 This object represents several ``<input type=radio>`` elements 536 that have the same name. 537 538 You can use this like a list, but also use the property 539 ``.value`` to check/uncheck inputs. Also you can use 540 ``.value_options`` to get the possible values. 541 c C s& | D ]}d|j v r|�d� S qdS )zs 542 Get/set the value, which checks the radio with that value (and 543 unchecks any other value). 544 r0 rP N�rs rC )rE r� r$ r$ r% rP � s 545 546 �zRadioGroup.valuec C sp d }|d ur | D ] }|� d�|kr|} nqtd| ��| D ] }d|jv r)|jd= q |d ur6|�dd� d S d S )NrP z)There is no radio input with the value %rr0 rA )rC rL rs rp )rE rP rx r� r$ r$ r% rP � s � 547 ��c C s 548 d | _ d S r] rU ra r$ r$ r% rP � s 549 c C � dd� | D �S )�< 550 Returns a list of all the possible values. 551 c S � g | ]}|� d ��qS rU rd r� r$ r$ r% rV � � z,RadioGroup.value_options.<locals>.<listcomp>r$ ra r$ r$ r% rz � r{ zRadioGroup.value_optionsc C � d| j jt�| �f S �Nz%s(%s)�r\ rj r� rV ra r$ r$ r% rV � rW zRadioGroup.__repr__N� 552 rj rk rl rm r� rP r� r� rz rV r$ r$ r$ r% r\ � s 553 554 555 556 557 r\ c @ r� ) r] aS 558 Represents a group of checkboxes (``<input type=checkbox>``) that 559 have the same name. 560 561 In addition to using this like a list, the ``.value`` attribute 562 returns a set-like object that you can add to or remove from to 563 check and uncheck checkboxes. You can also use ``.value_options`` 564 to get the possible values. 565 c C rx )z� 566 Return a set-like object that can be modified to check or 567 uncheck individual checkboxes according to their value. 568 )�CheckboxValuesra r$ r$ r% rP s zCheckboxGroup.valuec C s< | j }|�� t|d�std| d j|f ��|�|� d S )Nrb z<A CheckboxGroup (name=%r) must be set to a sequence (not %r)r )rP rw rA rL rT rg )rE rP re r$ r$ r% rP s 569 ��c C s | j �� d S r] )rP rw ra r$ r$ r% rP s c C r� )r� c S r� rU rd r� r$ r$ r% rV ) r� z/CheckboxGroup.value_options.<locals>.<listcomp>r$ ra r$ r$ r% rz $ r{ zCheckboxGroup.value_optionsc C r� r� r� ra r$ r$ r% rV + s �zCheckboxGroup.__repr__Nr� r$ r$ r$ r% r] s 570 571 572 573 r] c @ s8 e Zd ZdZdd� Zdd� Zdd� Zdd � Zd 574 d � Zd S ) r� zj 575 Represents the values of the checked checkboxes in a group of 576 checkboxes with the same name. 577 c C rM r] )r� )rE r� r$ r$ r% rG 6 r� zCheckboxValues.__init__c C s t dd� | jD ��S )Nc S s g | ] }d |j v r|�d��qS )r0 rP r� r� r$ r$ r% rV : s 578 579 �z+CheckboxValues.__iter__.<locals>.<listcomp>)r` r� ra r$ r$ r% rb 9 s �zCheckboxValues.__iter__c C s8 | j D ]}|�d�|kr|�dd� d S qtd| ��)NrP r0 rA �No checkbox with value %r)r� rC rp r� �rE rP r� r$ r$ r% rR ? s 580 � zCheckboxValues.addc C sJ | j D ]}|�d�|kr d|jv r|jd= d S td| ��qtd| ��)NrP r0 z0The checkbox with value %r was already uncheckedr� )r� rC rs r� r� r$ r$ r% rY G s 581 582 ����zCheckboxValues.removec C r~ )Nz <%s {%s} for checkboxes name=%r>r c S r r$ r r� r$ r$ r% rV W r z+CheckboxValues.__repr__.<locals>.<listcomp>)r\ rj rN r� rT ra r$ r$ r% rV T r� zCheckboxValues.__repr__N) rj rk rl rm rG rb rR rY rV r$ r$ r$ r% r� 0 s r� c @ sx e Zd ZdZedd� �Zejdd� �Zejdd� �Zedd� �Zejdd� �Zed d 583 � �Z ed d � �Z 584 e 585 jd d � �Z 586 dS )� InputElementaZ 587 Represents an ``<input>`` element. 588 589 You can get the type with ``.type`` (which is lower-cased and 590 defaults to ``'text'``). 591 592 Also you can get and set the value with ``.value`` 593 594 Checkboxes and radios have the attribute ``input.checkable == 595 True`` (for all others it is false) and a boolean attribute 596 ``.checked``. 597 598 c C s( | j r| jr | �d�p dS dS | �d�S )z� 599 Get/set the value of this element, using the ``value`` attribute. 600 601 Also, if this is a checkbox and it has no value, this defaults 602 to ``'on'``. If it is a checkbox or radio that is not 603 checked, this returns None. 604 rP �onN)r/ r0 rC ra r$ r$ r% rP k s 605 606 zInputElement.valuec C sH | j r |s 607 d| _d S d| _t|t�r| �d|� d S d S | �d|� d S )NFTrP )r/ r0 r8 r9 rp rZ r$ r$ r% rP { s 608 609 �c C s* | j rd| _d S d| jv r| jd= d S d S )NFrP )r/ r0 rs ra r$ r$ r% rP � s 610 611 612 �c C r= )zM 613 Return the type of this element (using the type attribute). 614 r� r� r� ra r$ r$ r% r� � s zInputElement.typec C r5 )Nr� r7 rZ r$ r$ r% r� � r8 c C s 615 | j dv S )z7 616 Boolean: can this element be checked? 617 )r[ rZ )r� ra r$ r$ r% r/ � rt zInputElement.checkablec C s | j std��d| jv S )z� 618 Boolean attribute to get/set the presence of the ``checked`` 619 attribute. 620 621 You can only use this on checkable input types. 622 �Not a checkable input typer0 )r/ �AttributeErrorrs ra r$ r$ r% r0 � s 623 zInputElement.checkedc C s> | j std��|r| �dd� d S | j}d|v r |d= d S d S )Nr� r0 rA )r/ r� rp rs )rE rP rs r$ r$ r% r0 � s 624 �N) rj rk rl rm r� rP r� r� r� r/ r0 r$ r$ r$ r% r� [ s$ 625 626 627 628 629 630 631 r� r) c @ rg )� LabelElementz� 632 Represents a ``<label>`` element. 633 634 Label elements are linked to other elements with their ``for`` 635 attribute. You can access this element with ``label.for_element``. 636 c C s | � d�}|s dS | j�|�S )zf 637 Get/set the element this label points to. Return None if it 638 can't be found. 639 r� N)rC r~ r� )rE r� r$ r$ r% � for_element� s 640 zLabelElement.for_elementc C s* |� d�}|s td| ��| �d|� d S )Nr� z Element %r has no id attributer� )rC r� rp )rE �otherr� r$ r$ r% r� � s 641 �c C r9 )Nr� r: r; r$ r$ r% r� � r< N)rj rk rl rm r� r� r� r� r$ r$ r$ r% r� � s 642 643 644 r� r� c C sZ z| � � } W n ty Y nw dt }| �tj�D ]}|j}|d dkr*|| |_qdS )zYConvert all tags in an HTML tree to XHTML by moving them to the 645 XHTML namespace. 646 ri r r6 N)�getrootr� r; r` r r r= )r� �prefixr� r= r$ r$ r% � html_to_xhtml� s � 647 ��r� c C sX z| � � } W n ty Y nw dt }t|�}| �|d �D ] 648 }|j|d� |_q dS )zUConvert all tags in an XHTML tree to HTML by removing their 649 XHTML namespace. 650 ri r� N)r� r� r; r: r` r= )�xhtmlr� � 651 prefix_lenr� r$ r$ r% � xhtml_to_html� s ��r� z%<meta http-equiv="Content-Type"[^>]*>rH r� c C sJ t j| |||||d�}|dkr#|s#t|t�r td|�}|S tt� |�}|S )a$ 652 Return an HTML string representation of the document. 653 654 Note: if include_meta_content_type is true this will create a 655 ``<meta http-equiv="Content-Type" ...>`` tag in the head; 656 regardless of the value of include_meta_content_type any existing 657 ``<meta http-equiv="Content-Type" ...>`` tag will be removed 658 659 The ``encoding`` argument controls the output encoding (defaults to 660 ASCII, with &#...; character references for any characters outside 661 of ASCII). Note that you can pass the name ``'unicode'`` as 662 ``encoding`` argument to serialise to a Unicode string. 663 664 The ``method`` argument defines the output method. It defaults to 665 'html', but can also be 'xml' for xhtml output, or 'text' to 666 serialise to plain text without markup. 667 668 To leave out the tail text of the top-level element that is being 669 serialised, pass ``with_tail=False``. 670 671 The ``doctype`` option allows passing in a plain string that will 672 be serialised before the XML tree. Note that passing in non 673 well-formed content here will make the XML output non well-formed. 674 Also, an existing doctype in the document tree will not be removed 675 when serialising an ElementTree instance. 676 677 Example:: 678 679 >>> from lxml import html 680 >>> root = html.fragment_fromstring('<p>Hello<br>world!</p>') 681 682 >>> html.tostring(root) 683 b'<p>Hello<br>world!</p>' 684 >>> html.tostring(root, method='html') 685 b'<p>Hello<br>world!</p>' 686 687 >>> html.tostring(root, method='xml') 688 b'<p>Hello<br/>world!</p>' 689 690 >>> html.tostring(root, method='text') 691 b'Helloworld!' 692 693 >>> html.tostring(root, method='text', encoding='unicode') 694 u'Helloworld!' 695 696 >>> root = html.fragment_fromstring('<div><p>Hello<br>world!</p>TAIL</div>') 697 >>> html.tostring(root[0], method='text', encoding='unicode') 698 u'Helloworld!TAIL' 699 700 >>> html.tostring(root[0], method='text', encoding='unicode', with_tail=False) 701 u'Helloworld!' 702 703 >>> doc = html.document_fromstring('<p>Hello<br>world!</p>') 704 >>> html.tostring(doc, method='html', encoding='unicode') 705 u'<html><body><p>Hello<br>world!</p></body></html>' 706 707 >>> print(html.tostring(doc, method='html', encoding='unicode', 708 ... doctype='<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"' 709 ... ' "http://www.w3.org/TR/html4/strict.dtd">')) 710 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd"> 711 <html><body><p>Hello<br>world!</p></body></html> 712 )r> � pretty_printr/ � with_tail�doctyper� rA )r r r8 r$ � __str_replace_meta_content_type�!__bytes_replace_meta_content_typer2 )r� r� Zinclude_meta_content_typer/ r> r� r� r� r$ r$ r% r s 713 ?� 714 715 �r c C s� ddl }ddl}ddl}t| tj�st�| �} |jdd�\}}|�|d�}z| j |d|p1| j 716 j p1dd� W |� � n|� � w d |� |jjd 717 � }t|� |�|� dS ) z� 718 Open the HTML document in a web browser, saving it to a temporary 719 file to open it. Note that this does not delete the file after 720 use. This is mainly meant for debugging. 721 r Nz.html)�suffix�wbr� zUTF-8rk zfile://�/)�os� 722 webbrowser�tempfiler8 r Z _ElementTree� ElementTree�mkstemp�fdopen�writeru r/ �close�replace�path�sep�print�open) r� r/ r� r� r� �handle�fn�fr� r$ r$ r% r 723 V s 724 r 725 c � e Zd ZdZ� fdd�Z� ZS )� 726 HTMLParserzOAn HTML parser that is configured to return lxml.html Element 727 objects. 728 c �& t t| �jdi |�� | �t� � d S �Nr$ )rX r� rG �set_element_class_lookupr� �rE �kwargsr[ r$ r% rG u � zHTMLParser.__init__�rj rk rl rm rG rn r$ r$ r[ r% r� q s r� c r� )� XHTMLParsera( An XML parser that is configured to return lxml.html Element 729 objects. 730 731 Note that this parser is not really XHTML aware unless you let it 732 load a DTD that declares the HTML entities. To do this, make sure 733 you have the XHTML DTDs installed in your catalogs, and create the 734 parser like this:: 735 736 >>> parser = XHTMLParser(load_dtd=True) 737 738 If you additionally want to validate the document, use this:: 739 740 >>> parser = XHTMLParser(dtd_validation=True) 741 742 For catalog support, see http://www.xmlsoft.org/catalog.html. 743 c r� r� )rX r� rG r� r� r� r[ r$ r% rG � r� zXHTMLParser.__init__r� r$ r$ r[ r% r� z s r� c O s t j| i |��}|S )zOCreate a new HTML Element. 744 745 This can also be used for XHTML documents. 746 )r� � makeelement)r� r� r2 r$ r$ r% r � s r )NF)FNNr� )FFNr� TNr] )lrm � 747 __future__r �__all__r� r r � functoolsr �collections.abcr r rJ � collectionsrA r r Z _setmixinr �urlparser rL r0 � NameErrorr$ r9 r2 r& r; �XPathr� rr ry r� r� r� r �Ir� r� r� r� r� rK r� r. r5 r> r? r� ro r� r r r r r r r r� Z ElementBaser� ZPIBaser� Z 748 EntityBaser� r� r� r� r r r r r r r r r r r r� r rB r r rc rh rn rq r� r\ r] r� r� r� r� r� r" r� r� r r 749 r� � XMLParserr� r r� � xhtml_parserr$ r$ r$ r% �<module> s � � � � 750 ��� 751 752 753 � 754 �� l $ . 755 �� �� 756 757 �' 758 � 759 1 760 D 761 t 762 &Y 763 764 d=5,+ 765 ^ 766 !���� 767 � J 768 769