<?xml version="1.0" encoding="ascii"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<title>lxml.html.soupparser</title>
<link rel="stylesheet" href="epydoc.css" type="text/css" />
<script type="text/javascript" src="epydoc.js"></script>
</head>
<body bgcolor="white" text="black" link="blue" vlink="#204080"
alink="#204080">
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Home link -->
<th> <a
href="lxml-module.html">Home</a> </th>
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<!-- Project homepage -->
<th class="navbar" align="right" width="100%">
<table border="0" cellpadding="0" cellspacing="0">
<tr><th class="navbar" align="center"
><a class="navbar" target="_top" href="/">lxml API</a></th>
</tr></table></th>
</tr>
</table>
<table width="100%" cellpadding="0" cellspacing="0">
<tr valign="top">
<td width="100%">
<span class="breadcrumbs">
<a href="lxml-module.html">Package lxml</a> ::
<a href="lxml.html-module.html">Package html</a> ::
Module soupparser
</span>
</td>
<td>
<table cellpadding="0" cellspacing="0">
<!-- hide/show private -->
<tr><td align="right"><span class="options">[<a href="javascript:void(0);" class="privatelink"
onclick="toggle_private();">hide private</a>]</span></td></tr>
<tr><td align="right"><span class="options"
>[<a href="frames.html" target="_top">frames</a
>] | <a href="lxml.html.soupparser-pysrc.html"
target="_top">no frames</a>]</span></td></tr>
</table>
</td>
</tr>
</table>
<h1 class="epydoc">Source Code for <a href="lxml.html.soupparser-module.html">Module lxml.html.soupparser</a></h1>
<pre class="py-src">
<a name="L1"></a><tt class="py-lineno"> 1</tt> <tt class="py-line"><tt class="py-docstring">"""External interface to the BeautifulSoup HTML parser.</tt> </tt>
<a name="L2"></a><tt class="py-lineno"> 2</tt> <tt class="py-line"><tt class="py-docstring">"""</tt> </tt>
<a name="L3"></a><tt class="py-lineno"> 3</tt> <tt class="py-line"> </tt>
<a name="L4"></a><tt class="py-lineno"> 4</tt> <tt class="py-line"><tt class="py-name">__all__</tt> <tt class="py-op">=</tt> <tt class="py-op">[</tt><tt class="py-string">"fromstring"</tt><tt class="py-op">,</tt> <tt class="py-string">"parse"</tt><tt class="py-op">,</tt> <tt class="py-string">"convert_tree"</tt><tt class="py-op">]</tt> </tt>
<a name="L5"></a><tt class="py-lineno"> 5</tt> <tt class="py-line"> </tt>
<a name="L6"></a><tt class="py-lineno"> 6</tt> <tt class="py-line"><tt class="py-keyword">import</tt> <tt class="py-name">re</tt> </tt>
<a name="L7"></a><tt class="py-lineno"> 7</tt> <tt class="py-line"><tt class="py-keyword">from</tt> <tt id="link-0" class="py-name" targets="Package lxml=lxml-module.html"><a title="lxml" class="py-name" href="#" onclick="return doclink('link-0', 'lxml', 'link-0');">lxml</a></tt> <tt class="py-keyword">import</tt> <tt id="link-1" class="py-name" targets="Module lxml.etree=lxml.etree-module.html,Variable lxml.sax.ElementTreeContentHandler.etree=lxml.sax.ElementTreeContentHandler-class.html#etree,Variable lxml.tests.test_elementtree.CElementTreeTestCase.etree=lxml.tests.test_elementtree.CElementTreeTestCase-class.html#etree,Variable lxml.tests.test_elementtree._ETreeTestCaseBase.etree=lxml.tests.test_elementtree._ETreeTestCaseBase-class.html#etree,Variable lxml.tests.test_elementtree._XMLPullParserTest.etree=lxml.tests.test_elementtree._XMLPullParserTest-class.html#etree,Variable lxml.tests.test_io._IOTestCaseBase.etree=lxml.tests.test_io._IOTestCaseBase-class.html#etree"><a title="lxml.etree
lxml.sax.ElementTreeContentHandler.etree
lxml.tests.test_elementtree.CElementTreeTestCase.etree
lxml.tests.test_elementtree._ETreeTestCaseBase.etree
lxml.tests.test_elementtree._XMLPullParserTest.etree
lxml.tests.test_io._IOTestCaseBase.etree" class="py-name" href="#" onclick="return doclink('link-1', 'etree', 'link-1');">etree</a></tt><tt class="py-op">,</tt> <tt id="link-2" class="py-name" targets="Package lxml.html=lxml.html-module.html,Method lxml.html.diff.href_token.html()=lxml.html.diff.href_token-class.html#html,Method lxml.html.diff.tag_token.html()=lxml.html.diff.tag_token-class.html#html,Method lxml.html.diff.token.html()=lxml.html.diff.token-class.html#html"><a title="lxml.html
lxml.html.diff.href_token.html
lxml.html.diff.tag_token.html
lxml.html.diff.token.html" class="py-name" href="#" onclick="return doclink('link-2', 'html', 'link-2');">html</a></tt> </tt>
<a name="L8"></a><tt class="py-lineno"> 8</tt> <tt class="py-line"> </tt>
<a name="L9"></a><tt class="py-lineno"> 9</tt> <tt class="py-line"><tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L10"></a><tt class="py-lineno"> 10</tt> <tt class="py-line"> <tt class="py-keyword">from</tt> <tt class="py-name">bs4</tt> <tt class="py-keyword">import</tt> <tt class="py-op">(</tt> </tt>
<a name="L11"></a><tt class="py-lineno"> 11</tt> <tt class="py-line"> <tt class="py-name">BeautifulSoup</tt><tt class="py-op">,</tt> <tt class="py-name">Tag</tt><tt class="py-op">,</tt> <tt id="link-3" class="py-name" targets="Function lxml.etree.Comment()=lxml.etree-module.html#Comment"><a title="lxml.etree.Comment" class="py-name" href="#" onclick="return doclink('link-3', 'Comment', 'link-3');">Comment</a></tt><tt class="py-op">,</tt> <tt class="py-name">ProcessingInstruction</tt><tt class="py-op">,</tt> <tt class="py-name">NavigableString</tt><tt class="py-op">,</tt> </tt>
<a name="L12"></a><tt class="py-lineno"> 12</tt> <tt class="py-line"> <tt class="py-name">Declaration</tt><tt class="py-op">,</tt> <tt class="py-name">Doctype</tt><tt class="py-op">)</tt> </tt>
<a name="L13"></a><tt class="py-lineno"> 13</tt> <tt class="py-line"> <tt id="link-4" class="py-name" targets="Variable lxml.html.soupparser._DECLARATION_OR_DOCTYPE=lxml.html.soupparser-module.html#_DECLARATION_OR_DOCTYPE"><a title="lxml.html.soupparser._DECLARATION_OR_DOCTYPE" class="py-name" href="#" onclick="return doclink('link-4', '_DECLARATION_OR_DOCTYPE', 'link-4');">_DECLARATION_OR_DOCTYPE</a></tt> <tt class="py-op">=</tt> <tt class="py-op">(</tt><tt class="py-name">Declaration</tt><tt class="py-op">,</tt> <tt class="py-name">Doctype</tt><tt class="py-op">)</tt> </tt>
<a name="L14"></a><tt class="py-lineno"> 14</tt> <tt class="py-line"><tt class="py-keyword">except</tt> <tt class="py-name">ImportError</tt><tt class="py-op">:</tt> </tt>
<a name="L15"></a><tt class="py-lineno"> 15</tt> <tt class="py-line"> <tt class="py-keyword">from</tt> <tt class="py-name">BeautifulSoup</tt> <tt class="py-keyword">import</tt> <tt class="py-op">(</tt> </tt>
<a name="L16"></a><tt class="py-lineno"> 16</tt> <tt class="py-line"> <tt class="py-name">BeautifulSoup</tt><tt class="py-op">,</tt> <tt class="py-name">Tag</tt><tt class="py-op">,</tt> <tt id="link-5" class="py-name"><a title="lxml.etree.Comment" class="py-name" href="#" onclick="return doclink('link-5', 'Comment', 'link-3');">Comment</a></tt><tt class="py-op">,</tt> <tt class="py-name">ProcessingInstruction</tt><tt class="py-op">,</tt> <tt class="py-name">NavigableString</tt><tt class="py-op">,</tt> </tt>
<a name="L17"></a><tt class="py-lineno"> 17</tt> <tt class="py-line"> <tt class="py-name">Declaration</tt><tt class="py-op">)</tt> </tt>
<a name="L18"></a><tt class="py-lineno"> 18</tt> <tt class="py-line"> <tt id="link-6" class="py-name"><a title="lxml.html.soupparser._DECLARATION_OR_DOCTYPE" class="py-name" href="#" onclick="return doclink('link-6', '_DECLARATION_OR_DOCTYPE', 'link-4');">_DECLARATION_OR_DOCTYPE</a></tt> <tt class="py-op">=</tt> <tt class="py-name">Declaration</tt> </tt>
<a name="fromstring"></a><div id="fromstring-def"><a name="L19"></a><tt class="py-lineno"> 19</tt> <tt class="py-line"> </tt>
<a name="L20"></a><tt class="py-lineno"> 20</tt> <tt class="py-line"> </tt>
<a name="L21"></a><tt class="py-lineno"> 21</tt> <a class="py-toggle" href="#" id="fromstring-toggle" onclick="return toggle('fromstring');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#fromstring">fromstring</a><tt class="py-op">(</tt><tt class="py-param">data</tt><tt class="py-op">,</tt> <tt class="py-param">beautifulsoup</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">,</tt> <tt class="py-param">makeelement</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">,</tt> <tt class="py-op">**</tt><tt class="py-param">bsargs</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="fromstring-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="fromstring-expanded"><a name="L22"></a><tt class="py-lineno"> 22</tt> <tt class="py-line"> <tt class="py-docstring">"""Parse a string of HTML data into an Element tree using the</tt> </tt>
<a name="L23"></a><tt class="py-lineno"> 23</tt> <tt class="py-line"><tt class="py-docstring"> BeautifulSoup parser.</tt> </tt>
<a name="L24"></a><tt class="py-lineno"> 24</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L25"></a><tt class="py-lineno"> 25</tt> <tt class="py-line"><tt class="py-docstring"> Returns the root ``<html>`` Element of the tree.</tt> </tt>
<a name="L26"></a><tt class="py-lineno"> 26</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L27"></a><tt class="py-lineno"> 27</tt> <tt class="py-line"><tt class="py-docstring"> You can pass a different BeautifulSoup parser through the</tt> </tt>
<a name="L28"></a><tt class="py-lineno"> 28</tt> <tt class="py-line"><tt class="py-docstring"> `beautifulsoup` keyword, and a diffent Element factory function</tt> </tt>
<a name="L29"></a><tt class="py-lineno"> 29</tt> <tt class="py-line"><tt class="py-docstring"> through the `makeelement` keyword. By default, the standard</tt> </tt>
<a name="L30"></a><tt class="py-lineno"> 30</tt> <tt class="py-line"><tt class="py-docstring"> ``BeautifulSoup`` class and the default factory of `lxml.html` are</tt> </tt>
<a name="L31"></a><tt class="py-lineno"> 31</tt> <tt class="py-line"><tt class="py-docstring"> used.</tt> </tt>
<a name="L32"></a><tt class="py-lineno"> 32</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="L33"></a><tt class="py-lineno"> 33</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-7" class="py-name" targets="Function lxml.html.soupparser._parse()=lxml.html.soupparser-module.html#_parse"><a title="lxml.html.soupparser._parse" class="py-name" href="#" onclick="return doclink('link-7', '_parse', 'link-7');">_parse</a></tt><tt class="py-op">(</tt><tt id="link-8" class="py-name" targets="Method lxml.etree.TreeBuilder.data()=lxml.etree.TreeBuilder-class.html#data"><a title="lxml.etree.TreeBuilder.data" class="py-name" href="#" onclick="return doclink('link-8', 'data', 'link-8');">data</a></tt><tt class="py-op">,</tt> <tt class="py-name">beautifulsoup</tt><tt class="py-op">,</tt> <tt id="link-9" class="py-name" targets="Method lxml.etree._Element.makeelement()=lxml.etree._Element-class.html#makeelement,Method lxml.etree.iterparse.makeelement()=lxml.etree.iterparse-class.html#makeelement,Function lxml.tests.selftest2.makeelement()=lxml.tests.selftest2-module.html#makeelement"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-9', 'makeelement', 'link-9');">makeelement</a></tt><tt class="py-op">,</tt> <tt class="py-op">**</tt><tt class="py-name">bsargs</tt><tt class="py-op">)</tt> </tt>
</div><a name="L34"></a><tt class="py-lineno"> 34</tt> <tt class="py-line"> </tt>
<a name="parse"></a><div id="parse-def"><a name="L35"></a><tt class="py-lineno"> 35</tt> <tt class="py-line"> </tt>
<a name="L36"></a><tt class="py-lineno"> 36</tt> <a class="py-toggle" href="#" id="parse-toggle" onclick="return toggle('parse');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#parse">parse</a><tt class="py-op">(</tt><tt class="py-param">file</tt><tt class="py-op">,</tt> <tt class="py-param">beautifulsoup</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">,</tt> <tt class="py-param">makeelement</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">,</tt> <tt class="py-op">**</tt><tt class="py-param">bsargs</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="parse-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="parse-expanded"><a name="L37"></a><tt class="py-lineno"> 37</tt> <tt class="py-line"> <tt class="py-docstring">"""Parse a file into an ElemenTree using the BeautifulSoup parser.</tt> </tt>
<a name="L38"></a><tt class="py-lineno"> 38</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L39"></a><tt class="py-lineno"> 39</tt> <tt class="py-line"><tt class="py-docstring"> You can pass a different BeautifulSoup parser through the</tt> </tt>
<a name="L40"></a><tt class="py-lineno"> 40</tt> <tt class="py-line"><tt class="py-docstring"> `beautifulsoup` keyword, and a diffent Element factory function</tt> </tt>
<a name="L41"></a><tt class="py-lineno"> 41</tt> <tt class="py-line"><tt class="py-docstring"> through the `makeelement` keyword. By default, the standard</tt> </tt>
<a name="L42"></a><tt class="py-lineno"> 42</tt> <tt class="py-line"><tt class="py-docstring"> ``BeautifulSoup`` class and the default factory of `lxml.html` are</tt> </tt>
<a name="L43"></a><tt class="py-lineno"> 43</tt> <tt class="py-line"><tt class="py-docstring"> used.</tt> </tt>
<a name="L44"></a><tt class="py-lineno"> 44</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="L45"></a><tt class="py-lineno"> 45</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-keyword">not</tt> <tt id="link-10" class="py-name" targets="Method lxml.objectify.ObjectPath.hasattr()=lxml.objectify.ObjectPath-class.html#hasattr"><a title="lxml.objectify.ObjectPath.hasattr" class="py-name" href="#" onclick="return doclink('link-10', 'hasattr', 'link-10');">hasattr</a></tt><tt class="py-op">(</tt><tt class="py-name">file</tt><tt class="py-op">,</tt> <tt class="py-string">'read'</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L46"></a><tt class="py-lineno"> 46</tt> <tt class="py-line"> <tt class="py-name">file</tt> <tt class="py-op">=</tt> <tt class="py-name">open</tt><tt class="py-op">(</tt><tt class="py-name">file</tt><tt class="py-op">)</tt> </tt>
<a name="L47"></a><tt class="py-lineno"> 47</tt> <tt class="py-line"> <tt id="link-11" class="py-name" targets="Variable lxml.etree.iterparse.root=lxml.etree.iterparse-class.html#root"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-11', 'root', 'link-11');">root</a></tt> <tt class="py-op">=</tt> <tt id="link-12" class="py-name"><a title="lxml.html.soupparser._parse" class="py-name" href="#" onclick="return doclink('link-12', '_parse', 'link-7');">_parse</a></tt><tt class="py-op">(</tt><tt class="py-name">file</tt><tt class="py-op">,</tt> <tt class="py-name">beautifulsoup</tt><tt class="py-op">,</tt> <tt id="link-13" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-13', 'makeelement', 'link-9');">makeelement</a></tt><tt class="py-op">,</tt> <tt class="py-op">**</tt><tt class="py-name">bsargs</tt><tt class="py-op">)</tt> </tt>
<a name="L48"></a><tt class="py-lineno"> 48</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-14" class="py-name"><a title="lxml.etree
lxml.sax.ElementTreeContentHandler.etree
lxml.tests.test_elementtree.CElementTreeTestCase.etree
lxml.tests.test_elementtree._ETreeTestCaseBase.etree
lxml.tests.test_elementtree._XMLPullParserTest.etree
lxml.tests.test_io._IOTestCaseBase.etree" class="py-name" href="#" onclick="return doclink('link-14', 'etree', 'link-1');">etree</a></tt><tt class="py-op">.</tt><tt id="link-15" class="py-name" targets="Function lxml.etree.ElementTree()=lxml.etree-module.html#ElementTree,Module xml.etree.ElementTree=xml.etree.ElementTree-module.html,Class xml.etree.ElementTree.ElementTree=xml.etree.ElementTree.ElementTree-class.html"><a title="lxml.etree.ElementTree
xml.etree.ElementTree
xml.etree.ElementTree.ElementTree" class="py-name" href="#" onclick="return doclink('link-15', 'ElementTree', 'link-15');">ElementTree</a></tt><tt class="py-op">(</tt><tt id="link-16" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-16', 'root', 'link-11');">root</a></tt><tt class="py-op">)</tt> </tt>
</div><a name="L49"></a><tt class="py-lineno"> 49</tt> <tt class="py-line"> </tt>
<a name="convert_tree"></a><div id="convert_tree-def"><a name="L50"></a><tt class="py-lineno"> 50</tt> <tt class="py-line"> </tt>
<a name="L51"></a><tt class="py-lineno"> 51</tt> <a class="py-toggle" href="#" id="convert_tree-toggle" onclick="return toggle('convert_tree');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#convert_tree">convert_tree</a><tt class="py-op">(</tt><tt class="py-param">beautiful_soup_tree</tt><tt class="py-op">,</tt> <tt class="py-param">makeelement</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="convert_tree-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="convert_tree-expanded"><a name="L52"></a><tt class="py-lineno"> 52</tt> <tt class="py-line"> <tt class="py-docstring">"""Convert a BeautifulSoup tree to a list of Element trees.</tt> </tt>
<a name="L53"></a><tt class="py-lineno"> 53</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L54"></a><tt class="py-lineno"> 54</tt> <tt class="py-line"><tt class="py-docstring"> Returns a list instead of a single root Element to support</tt> </tt>
<a name="L55"></a><tt class="py-lineno"> 55</tt> <tt class="py-line"><tt class="py-docstring"> HTML-like soup with more than one root element.</tt> </tt>
<a name="L56"></a><tt class="py-lineno"> 56</tt> <tt class="py-line"><tt class="py-docstring"></tt> </tt>
<a name="L57"></a><tt class="py-lineno"> 57</tt> <tt class="py-line"><tt class="py-docstring"> You can pass a different Element factory through the `makeelement`</tt> </tt>
<a name="L58"></a><tt class="py-lineno"> 58</tt> <tt class="py-line"><tt class="py-docstring"> keyword.</tt> </tt>
<a name="L59"></a><tt class="py-lineno"> 59</tt> <tt class="py-line"><tt class="py-docstring"> """</tt> </tt>
<a name="L60"></a><tt class="py-lineno"> 60</tt> <tt class="py-line"> <tt id="link-17" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-17', 'root', 'link-11');">root</a></tt> <tt class="py-op">=</tt> <tt id="link-18" class="py-name" targets="Function lxml.html.soupparser._convert_tree()=lxml.html.soupparser-module.html#_convert_tree"><a title="lxml.html.soupparser._convert_tree" class="py-name" href="#" onclick="return doclink('link-18', '_convert_tree', 'link-18');">_convert_tree</a></tt><tt class="py-op">(</tt><tt class="py-name">beautiful_soup_tree</tt><tt class="py-op">,</tt> <tt id="link-19" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-19', 'makeelement', 'link-9');">makeelement</a></tt><tt class="py-op">)</tt> </tt>
<a name="L61"></a><tt class="py-lineno"> 61</tt> <tt class="py-line"> <tt class="py-name">children</tt> <tt class="py-op">=</tt> <tt id="link-20" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-20', 'root', 'link-11');">root</a></tt><tt class="py-op">.</tt><tt id="link-21" class="py-name" targets="Method lxml.etree._Element.getchildren()=lxml.etree._Element-class.html#getchildren,Method lxml.objectify.ObjectifiedElement.getchildren()=lxml.objectify.ObjectifiedElement-class.html#getchildren"><a title="lxml.etree._Element.getchildren
lxml.objectify.ObjectifiedElement.getchildren" class="py-name" href="#" onclick="return doclink('link-21', 'getchildren', 'link-21');">getchildren</a></tt><tt class="py-op">(</tt><tt class="py-op">)</tt> </tt>
<a name="L62"></a><tt class="py-lineno"> 62</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">child</tt> <tt class="py-keyword">in</tt> <tt class="py-name">children</tt><tt class="py-op">:</tt> </tt>
<a name="L63"></a><tt class="py-lineno"> 63</tt> <tt class="py-line"> <tt id="link-22" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-22', 'root', 'link-11');">root</a></tt><tt class="py-op">.</tt><tt id="link-23" class="py-name" targets="Method lxml.etree._Element.remove()=lxml.etree._Element-class.html#remove,Method lxml.html.CheckboxValues.remove()=lxml.html.CheckboxValues-class.html#remove,Method lxml.html.Classes.remove()=lxml.html.Classes-class.html#remove,Method lxml.html.MultipleSelectOptions.remove()=lxml.html.MultipleSelectOptions-class.html#remove"><a title="lxml.etree._Element.remove
lxml.html.CheckboxValues.remove
lxml.html.Classes.remove
lxml.html.MultipleSelectOptions.remove" class="py-name" href="#" onclick="return doclink('link-23', 'remove', 'link-23');">remove</a></tt><tt class="py-op">(</tt><tt class="py-name">child</tt><tt class="py-op">)</tt> </tt>
<a name="L64"></a><tt class="py-lineno"> 64</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">children</tt> </tt>
</div><a name="L65"></a><tt class="py-lineno"> 65</tt> <tt class="py-line"> </tt>
<a name="_parse"></a><div id="_parse-def"><a name="L66"></a><tt class="py-lineno"> 66</tt> <tt class="py-line"> </tt>
<a name="L67"></a><tt class="py-lineno"> 67</tt> <tt class="py-line"><tt class="py-comment"># helpers</tt> </tt>
<a name="L68"></a><tt class="py-lineno"> 68</tt> <tt class="py-line"> </tt>
<a name="L69"></a><tt class="py-lineno"> 69</tt> <a class="py-toggle" href="#" id="_parse-toggle" onclick="return toggle('_parse');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#_parse">_parse</a><tt class="py-op">(</tt><tt class="py-param">source</tt><tt class="py-op">,</tt> <tt class="py-param">beautifulsoup</tt><tt class="py-op">,</tt> <tt class="py-param">makeelement</tt><tt class="py-op">,</tt> <tt class="py-op">**</tt><tt class="py-param">bsargs</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="_parse-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="_parse-expanded"><a name="L70"></a><tt class="py-lineno"> 70</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">beautifulsoup</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L71"></a><tt class="py-lineno"> 71</tt> <tt class="py-line"> <tt class="py-name">beautifulsoup</tt> <tt class="py-op">=</tt> <tt class="py-name">BeautifulSoup</tt> </tt>
<a name="L72"></a><tt class="py-lineno"> 72</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt id="link-24" class="py-name"><a title="lxml.objectify.ObjectPath.hasattr" class="py-name" href="#" onclick="return doclink('link-24', 'hasattr', 'link-10');">hasattr</a></tt><tt class="py-op">(</tt><tt class="py-name">beautifulsoup</tt><tt class="py-op">,</tt> <tt class="py-string">"HTML_ENTITIES"</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> <tt class="py-comment"># bs3</tt> </tt>
<a name="L73"></a><tt class="py-lineno"> 73</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-string">'convertEntities'</tt> <tt class="py-keyword">not</tt> <tt class="py-keyword">in</tt> <tt class="py-name">bsargs</tt><tt class="py-op">:</tt> </tt>
<a name="L74"></a><tt class="py-lineno"> 74</tt> <tt class="py-line"> <tt class="py-name">bsargs</tt><tt class="py-op">[</tt><tt class="py-string">'convertEntities'</tt><tt class="py-op">]</tt> <tt class="py-op">=</tt> <tt class="py-string">'html'</tt> </tt>
<a name="L75"></a><tt class="py-lineno"> 75</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt id="link-25" class="py-name"><a title="lxml.objectify.ObjectPath.hasattr" class="py-name" href="#" onclick="return doclink('link-25', 'hasattr', 'link-10');">hasattr</a></tt><tt class="py-op">(</tt><tt class="py-name">beautifulsoup</tt><tt class="py-op">,</tt> <tt class="py-string">"DEFAULT_BUILDER_FEATURES"</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> <tt class="py-comment"># bs4</tt> </tt>
<a name="L76"></a><tt class="py-lineno"> 76</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-string">'features'</tt> <tt class="py-keyword">not</tt> <tt class="py-keyword">in</tt> <tt class="py-name">bsargs</tt><tt class="py-op">:</tt> </tt>
<a name="L77"></a><tt class="py-lineno"> 77</tt> <tt class="py-line"> <tt class="py-name">bsargs</tt><tt class="py-op">[</tt><tt class="py-string">'features'</tt><tt class="py-op">]</tt> <tt class="py-op">=</tt> <tt class="py-string">'html.parser'</tt> <tt class="py-comment"># use Python html parser</tt> </tt>
<a name="L78"></a><tt class="py-lineno"> 78</tt> <tt class="py-line"> <tt class="py-name">tree</tt> <tt class="py-op">=</tt> <tt class="py-name">beautifulsoup</tt><tt class="py-op">(</tt><tt class="py-name">source</tt><tt class="py-op">,</tt> <tt class="py-op">**</tt><tt class="py-name">bsargs</tt><tt class="py-op">)</tt> </tt>
<a name="L79"></a><tt class="py-lineno"> 79</tt> <tt class="py-line"> <tt id="link-26" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-26', 'root', 'link-11');">root</a></tt> <tt class="py-op">=</tt> <tt id="link-27" class="py-name"><a title="lxml.html.soupparser._convert_tree" class="py-name" href="#" onclick="return doclink('link-27', '_convert_tree', 'link-18');">_convert_tree</a></tt><tt class="py-op">(</tt><tt class="py-name">tree</tt><tt class="py-op">,</tt> <tt id="link-28" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-28', 'makeelement', 'link-9');">makeelement</a></tt><tt class="py-op">)</tt> </tt>
<a name="L80"></a><tt class="py-lineno"> 80</tt> <tt class="py-line"> <tt class="py-comment"># from ET: wrap the document in a html root element, if necessary</tt> </tt>
<a name="L81"></a><tt class="py-lineno"> 81</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">len</tt><tt class="py-op">(</tt><tt id="link-29" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-29', 'root', 'link-11');">root</a></tt><tt class="py-op">)</tt> <tt class="py-op">==</tt> <tt class="py-number">1</tt> <tt class="py-keyword">and</tt> <tt id="link-30" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-30', 'root', 'link-11');">root</a></tt><tt class="py-op">[</tt><tt class="py-number">0</tt><tt class="py-op">]</tt><tt class="py-op">.</tt><tt id="link-31" class="py-name" targets="Variable lxml.etree._Comment.tag=lxml.etree._Comment-class.html#tag,Variable lxml.etree._Element.tag=lxml.etree._Element-class.html#tag,Variable lxml.etree._Entity.tag=lxml.etree._Entity-class.html#tag,Variable lxml.etree._ProcessingInstruction.tag=lxml.etree._ProcessingInstruction-class.html#tag,Function lxml.tests.test_xpathevaluator.tag()=lxml.tests.test_xpathevaluator-module.html#tag,Variable xml.etree.ElementTree.Element.tag=xml.etree.ElementTree.Element-class.html#tag"><a title="lxml.etree._Comment.tag
lxml.etree._Element.tag
lxml.etree._Entity.tag
lxml.etree._ProcessingInstruction.tag
lxml.tests.test_xpathevaluator.tag
xml.etree.ElementTree.Element.tag" class="py-name" href="#" onclick="return doclink('link-31', 'tag', 'link-31');">tag</a></tt> <tt class="py-op">==</tt> <tt class="py-string">"html"</tt><tt class="py-op">:</tt> </tt>
<a name="L82"></a><tt class="py-lineno"> 82</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-32" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-32', 'root', 'link-11');">root</a></tt><tt class="py-op">[</tt><tt class="py-number">0</tt><tt class="py-op">]</tt> </tt>
<a name="L83"></a><tt class="py-lineno"> 83</tt> <tt class="py-line"> <tt id="link-33" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-33', 'root', 'link-11');">root</a></tt><tt class="py-op">.</tt><tt id="link-34" class="py-name"><a title="lxml.etree._Comment.tag
lxml.etree._Element.tag
lxml.etree._Entity.tag
lxml.etree._ProcessingInstruction.tag
lxml.tests.test_xpathevaluator.tag
xml.etree.ElementTree.Element.tag" class="py-name" href="#" onclick="return doclink('link-34', 'tag', 'link-31');">tag</a></tt> <tt class="py-op">=</tt> <tt class="py-string">"html"</tt> </tt>
<a name="L84"></a><tt class="py-lineno"> 84</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-35" class="py-name"><a title="lxml.etree.iterparse.root" class="py-name" href="#" onclick="return doclink('link-35', 'root', 'link-11');">root</a></tt> </tt>
</div><a name="L85"></a><tt class="py-lineno"> 85</tt> <tt class="py-line"> </tt>
<a name="L86"></a><tt class="py-lineno"> 86</tt> <tt class="py-line"> </tt>
<a name="L87"></a><tt class="py-lineno"> 87</tt> <tt class="py-line"><tt id="link-36" class="py-name" targets="Function lxml.html.soupparser._parse_doctype_declaration()=lxml.html.soupparser-module.html#_parse_doctype_declaration"><a title="lxml.html.soupparser._parse_doctype_declaration" class="py-name" href="#" onclick="return doclink('link-36', '_parse_doctype_declaration', 'link-36');">_parse_doctype_declaration</a></tt> <tt class="py-op">=</tt> <tt class="py-name">re</tt><tt class="py-op">.</tt><tt class="py-name">compile</tt><tt class="py-op">(</tt> </tt>
<a name="L88"></a><tt class="py-lineno"> 88</tt> <tt class="py-line"> <tt class="py-string">r'(?:\s|[<!])*DOCTYPE\s*HTML'</tt> </tt>
<a name="L89"></a><tt class="py-lineno"> 89</tt> <tt class="py-line"> <tt class="py-string">r'(?:\s+PUBLIC)?(?:\s+(\'[^\']*\'|"[^"]*"))?'</tt> </tt>
<a name="L90"></a><tt class="py-lineno"> 90</tt> <tt class="py-line"> <tt class="py-string">r'(?:\s+(\'[^\']*\'|"[^"]*"))?'</tt><tt class="py-op">,</tt> </tt>
<a name="L91"></a><tt class="py-lineno"> 91</tt> <tt class="py-line"> <tt class="py-name">re</tt><tt class="py-op">.</tt><tt class="py-name">IGNORECASE</tt><tt class="py-op">)</tt><tt class="py-op">.</tt><tt class="py-name">match</tt> </tt>
<a name="_PseudoTag"></a><div id="_PseudoTag-def"><a name="L92"></a><tt class="py-lineno"> 92</tt> <tt class="py-line"> </tt>
<a name="L93"></a><tt class="py-lineno"> 93</tt> <tt class="py-line"> </tt>
<a name="L94"></a><tt class="py-lineno"> 94</tt> <a class="py-toggle" href="#" id="_PseudoTag-toggle" onclick="return toggle('_PseudoTag');">-</a><tt class="py-line"><tt class="py-keyword">class</tt> <a class="py-def-name" href="lxml.html.soupparser._PseudoTag-class.html">_PseudoTag</a><tt class="py-op">:</tt> </tt>
</div><div id="_PseudoTag-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="_PseudoTag-expanded"><a name="L95"></a><tt class="py-lineno"> 95</tt> <tt class="py-line"> <tt class="py-comment"># Minimal imitation of BeautifulSoup.Tag</tt> </tt>
<a name="_PseudoTag.__init__"></a><div id="_PseudoTag.__init__-def"><a name="L96"></a><tt class="py-lineno"> 96</tt> <a class="py-toggle" href="#" id="_PseudoTag.__init__-toggle" onclick="return toggle('_PseudoTag.__init__');">-</a><tt class="py-line"> <tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser._PseudoTag-class.html#__init__">__init__</a><tt class="py-op">(</tt><tt class="py-param">self</tt><tt class="py-op">,</tt> <tt class="py-param">contents</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="_PseudoTag.__init__-collapsed" style="display:none;" pad="+++" indent="++++++++"></div><div id="_PseudoTag.__init__-expanded"><a name="L97"></a><tt class="py-lineno"> 97</tt> <tt class="py-line"> <tt class="py-name">self</tt><tt class="py-op">.</tt><tt id="link-37" class="py-name" targets="Variable lxml.etree.DTD.name=lxml.etree.DTD-class.html#name,Variable lxml.etree._Entity.name=lxml.etree._Entity-class.html#name,Variable lxml.html.InputMixin.name=lxml.html.InputMixin-class.html#name,Variable lxml.objectify.PyType.name=lxml.objectify.PyType-class.html#name"><a title="lxml.etree.DTD.name
lxml.etree._Entity.name
lxml.html.InputMixin.name
lxml.objectify.PyType.name" class="py-name" href="#" onclick="return doclink('link-37', 'name', 'link-37');">name</a></tt> <tt class="py-op">=</tt> <tt class="py-string">'html'</tt> </tt>
<a name="L98"></a><tt class="py-lineno"> 98</tt> <tt class="py-line"> <tt class="py-name">self</tt><tt class="py-op">.</tt><tt class="py-name">attrs</tt> <tt class="py-op">=</tt> <tt class="py-op">[</tt><tt class="py-op">]</tt> </tt>
<a name="L99"></a><tt class="py-lineno"> 99</tt> <tt class="py-line"> <tt class="py-name">self</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt> <tt class="py-op">=</tt> <tt class="py-name">contents</tt> </tt>
</div><a name="L100"></a><tt class="py-lineno">100</tt> <tt class="py-line"> </tt>
<a name="_PseudoTag.__iter__"></a><div id="_PseudoTag.__iter__-def"><a name="L101"></a><tt class="py-lineno">101</tt> <a class="py-toggle" href="#" id="_PseudoTag.__iter__-toggle" onclick="return toggle('_PseudoTag.__iter__');">-</a><tt class="py-line"> <tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser._PseudoTag-class.html#__iter__">__iter__</a><tt class="py-op">(</tt><tt class="py-param">self</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="_PseudoTag.__iter__-collapsed" style="display:none;" pad="+++" indent="++++++++"></div><div id="_PseudoTag.__iter__-expanded"><a name="L102"></a><tt class="py-lineno">102</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">self</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt><tt class="py-op">.</tt><tt id="link-38" class="py-name" targets="Method lxml.etree.ElementDepthFirstIterator.__iter__()=lxml.etree.ElementDepthFirstIterator-class.html#__iter__,Method lxml.etree.ElementTextIterator.__iter__()=lxml.etree.ElementTextIterator-class.html#__iter__,Method lxml.etree._Attrib.__iter__()=lxml.etree._Attrib-class.html#__iter__,Method lxml.etree._Element.__iter__()=lxml.etree._Element-class.html#__iter__,Method lxml.etree._ElementIterator.__iter__()=lxml.etree._ElementIterator-class.html#__iter__,Method lxml.etree._ElementMatchIterator.__iter__()=lxml.etree._ElementMatchIterator-class.html#__iter__,Method lxml.etree._ErrorLog.__iter__()=lxml.etree._ErrorLog-class.html#__iter__,Method lxml.etree._IDDict.__iter__()=lxml.etree._IDDict-class.html#__iter__,Method lxml.etree._ListErrorLog.__iter__()=lxml.etree._ListErrorLog-class.html#__iter__,Method lxml.etree.iterparse.__iter__()=lxml.etree.iterparse-class.html#__iter__,Method lxml.etree.iterwalk.__iter__()=lxml.etree.iterwalk-class.html#__iter__,Method lxml.html.CheckboxValues.__iter__()=lxml.html.CheckboxValues-class.html#__iter__,Method lxml.html.Classes.__iter__()=lxml.html.Classes-class.html#__iter__,Method lxml.html.FieldsDict.__iter__()=lxml.html.FieldsDict-class.html#__iter__,Method lxml.html.InputGetter.__iter__()=lxml.html.InputGetter-class.html#__iter__,Method lxml.html.MultipleSelectOptions.__iter__()=lxml.html.MultipleSelectOptions-class.html#__iter__,Method lxml.html.soupparser._PseudoTag.__iter__()=lxml.html.soupparser._PseudoTag-class.html#__iter__,Method lxml.objectify.ObjectifiedElement.__iter__()=lxml.objectify.ObjectifiedElement-class.html#__iter__"><a title="lxml.etree.ElementDepthFirstIterator.__iter__
lxml.etree.ElementTextIterator.__iter__
lxml.etree._Attrib.__iter__
lxml.etree._Element.__iter__
lxml.etree._ElementIterator.__iter__
lxml.etree._ElementMatchIterator.__iter__
lxml.etree._ErrorLog.__iter__
lxml.etree._IDDict.__iter__
lxml.etree._ListErrorLog.__iter__
lxml.etree.iterparse.__iter__
lxml.etree.iterwalk.__iter__
lxml.html.CheckboxValues.__iter__
lxml.html.Classes.__iter__
lxml.html.FieldsDict.__iter__
lxml.html.InputGetter.__iter__
lxml.html.MultipleSelectOptions.__iter__
lxml.html.soupparser._PseudoTag.__iter__
lxml.objectify.ObjectifiedElement.__iter__" class="py-name" href="#" onclick="return doclink('link-38', '__iter__', 'link-38');">__iter__</a></tt><tt class="py-op">(</tt><tt class="py-op">)</tt> </tt>
</div></div><a name="L103"></a><tt class="py-lineno">103</tt> <tt class="py-line"> </tt>
<a name="_convert_tree"></a><div id="_convert_tree-def"><a name="L104"></a><tt class="py-lineno">104</tt> <tt class="py-line"> </tt>
<a name="L105"></a><tt class="py-lineno">105</tt> <a class="py-toggle" href="#" id="_convert_tree-toggle" onclick="return toggle('_convert_tree');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#_convert_tree">_convert_tree</a><tt class="py-op">(</tt><tt class="py-param">beautiful_soup_tree</tt><tt class="py-op">,</tt> <tt class="py-param">makeelement</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="_convert_tree-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="_convert_tree-expanded"><a name="L106"></a><tt class="py-lineno">106</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt id="link-39" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-39', 'makeelement', 'link-9');">makeelement</a></tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L107"></a><tt class="py-lineno">107</tt> <tt class="py-line"> <tt id="link-40" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-40', 'makeelement', 'link-9');">makeelement</a></tt> <tt class="py-op">=</tt> <tt id="link-41" class="py-name"><a title="lxml.html
lxml.html.diff.href_token.html
lxml.html.diff.tag_token.html
lxml.html.diff.token.html" class="py-name" href="#" onclick="return doclink('link-41', 'html', 'link-2');">html</a></tt><tt class="py-op">.</tt><tt id="link-42" class="py-name" targets="Variable lxml.html.html5parser.html_parser=lxml.html.html5parser-module.html#html_parser,Variable lxml.html.html_parser=lxml.html-module.html#html_parser"><a title="lxml.html.html5parser.html_parser
lxml.html.html_parser" class="py-name" href="#" onclick="return doclink('link-42', 'html_parser', 'link-42');">html_parser</a></tt><tt class="py-op">.</tt><tt id="link-43" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-43', 'makeelement', 'link-9');">makeelement</a></tt> </tt>
<a name="L108"></a><tt class="py-lineno">108</tt> <tt class="py-line"> </tt>
<a name="L109"></a><tt class="py-lineno">109</tt> <tt class="py-line"> <tt class="py-comment"># Split the tree into three parts:</tt> </tt>
<a name="L110"></a><tt class="py-lineno">110</tt> <tt class="py-line"> <tt class="py-comment"># i) everything before the root element: document type</tt> </tt>
<a name="L111"></a><tt class="py-lineno">111</tt> <tt class="py-line"> <tt class="py-comment"># declaration, comments, processing instructions, whitespace</tt> </tt>
<a name="L112"></a><tt class="py-lineno">112</tt> <tt class="py-line"> <tt class="py-comment"># ii) the root(s),</tt> </tt>
<a name="L113"></a><tt class="py-lineno">113</tt> <tt class="py-line"> <tt class="py-comment"># iii) everything after the root: comments, processing</tt> </tt>
<a name="L114"></a><tt class="py-lineno">114</tt> <tt class="py-line"> <tt class="py-comment"># instructions, whitespace</tt> </tt>
<a name="L115"></a><tt class="py-lineno">115</tt> <tt class="py-line"> <tt class="py-name">first_element_idx</tt> <tt class="py-op">=</tt> <tt class="py-name">last_element_idx</tt> <tt class="py-op">=</tt> <tt class="py-name">None</tt> </tt>
<a name="L116"></a><tt class="py-lineno">116</tt> <tt class="py-line"> <tt class="py-name">html_root</tt> <tt class="py-op">=</tt> <tt class="py-name">declaration</tt> <tt class="py-op">=</tt> <tt class="py-name">None</tt> </tt>
<a name="L117"></a><tt class="py-lineno">117</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">i</tt><tt class="py-op">,</tt> <tt class="py-name">e</tt> <tt class="py-keyword">in</tt> <tt class="py-name">enumerate</tt><tt class="py-op">(</tt><tt class="py-name">beautiful_soup_tree</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L118"></a><tt class="py-lineno">118</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">isinstance</tt><tt class="py-op">(</tt><tt class="py-name">e</tt><tt class="py-op">,</tt> <tt class="py-name">Tag</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L119"></a><tt class="py-lineno">119</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">first_element_idx</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L120"></a><tt class="py-lineno">120</tt> <tt class="py-line"> <tt class="py-name">first_element_idx</tt> <tt class="py-op">=</tt> <tt class="py-name">i</tt> </tt>
<a name="L121"></a><tt class="py-lineno">121</tt> <tt class="py-line"> <tt class="py-name">last_element_idx</tt> <tt class="py-op">=</tt> <tt class="py-name">i</tt> </tt>
<a name="L122"></a><tt class="py-lineno">122</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">html_root</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt> <tt class="py-keyword">and</tt> <tt class="py-name">e</tt><tt class="py-op">.</tt><tt id="link-44" class="py-name"><a title="lxml.etree.DTD.name
lxml.etree._Entity.name
lxml.html.InputMixin.name
lxml.objectify.PyType.name" class="py-name" href="#" onclick="return doclink('link-44', 'name', 'link-37');">name</a></tt> <tt class="py-keyword">and</tt> <tt class="py-name">e</tt><tt class="py-op">.</tt><tt id="link-45" class="py-name"><a title="lxml.etree.DTD.name
lxml.etree._Entity.name
lxml.html.InputMixin.name
lxml.objectify.PyType.name" class="py-name" href="#" onclick="return doclink('link-45', 'name', 'link-37');">name</a></tt><tt class="py-op">.</tt><tt class="py-name">lower</tt><tt class="py-op">(</tt><tt class="py-op">)</tt> <tt class="py-op">==</tt> <tt class="py-string">'html'</tt><tt class="py-op">:</tt> </tt>
<a name="L123"></a><tt class="py-lineno">123</tt> <tt class="py-line"> <tt class="py-name">html_root</tt> <tt class="py-op">=</tt> <tt class="py-name">e</tt> </tt>
<a name="L124"></a><tt class="py-lineno">124</tt> <tt class="py-line"> <tt class="py-keyword">elif</tt> <tt class="py-name">declaration</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt> <tt class="py-keyword">and</tt> <tt class="py-name">isinstance</tt><tt class="py-op">(</tt><tt class="py-name">e</tt><tt class="py-op">,</tt> <tt id="link-46" class="py-name"><a title="lxml.html.soupparser._DECLARATION_OR_DOCTYPE" class="py-name" href="#" onclick="return doclink('link-46', '_DECLARATION_OR_DOCTYPE', 'link-4');">_DECLARATION_OR_DOCTYPE</a></tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L125"></a><tt class="py-lineno">125</tt> <tt class="py-line"> <tt class="py-name">declaration</tt> <tt class="py-op">=</tt> <tt class="py-name">e</tt> </tt>
<a name="L126"></a><tt class="py-lineno">126</tt> <tt class="py-line"> </tt>
<a name="L127"></a><tt class="py-lineno">127</tt> <tt class="py-line"> <tt class="py-comment"># For a nice, well-formatted document, the variable roots below is</tt> </tt>
<a name="L128"></a><tt class="py-lineno">128</tt> <tt class="py-line"> <tt class="py-comment"># a list consisting of a single <html> element. However, the document</tt> </tt>
<a name="L129"></a><tt class="py-lineno">129</tt> <tt class="py-line"> <tt class="py-comment"># may be a soup like '<meta><head><title>Hello</head><body>Hi</tt> </tt>
<a name="L130"></a><tt class="py-lineno">130</tt> <tt class="py-line"> <tt class="py-comment"># all<\p>'. In this example roots is a list containing meta, head</tt> </tt>
<a name="L131"></a><tt class="py-lineno">131</tt> <tt class="py-line"> <tt class="py-comment"># and body elements.</tt> </tt>
<a name="L132"></a><tt class="py-lineno">132</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">first_element_idx</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L133"></a><tt class="py-lineno">133</tt> <tt class="py-line"> <tt class="py-name">pre_root</tt> <tt class="py-op">=</tt> <tt class="py-name">post_root</tt> <tt class="py-op">=</tt> <tt class="py-op">[</tt><tt class="py-op">]</tt> </tt>
<a name="L134"></a><tt class="py-lineno">134</tt> <tt class="py-line"> <tt class="py-name">roots</tt> <tt class="py-op">=</tt> <tt class="py-name">beautiful_soup_tree</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt> </tt>
<a name="L135"></a><tt class="py-lineno">135</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L136"></a><tt class="py-lineno">136</tt> <tt class="py-line"> <tt class="py-name">pre_root</tt> <tt class="py-op">=</tt> <tt class="py-name">beautiful_soup_tree</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt><tt class="py-op">[</tt><tt class="py-op">:</tt><tt class="py-name">first_element_idx</tt><tt class="py-op">]</tt> </tt>
<a name="L137"></a><tt class="py-lineno">137</tt> <tt class="py-line"> <tt class="py-name">roots</tt> <tt class="py-op">=</tt> <tt class="py-name">beautiful_soup_tree</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt><tt class="py-op">[</tt><tt class="py-name">first_element_idx</tt><tt class="py-op">:</tt><tt class="py-name">last_element_idx</tt><tt class="py-op">+</tt><tt class="py-number">1</tt><tt class="py-op">]</tt> </tt>
<a name="L138"></a><tt class="py-lineno">138</tt> <tt class="py-line"> <tt class="py-name">post_root</tt> <tt class="py-op">=</tt> <tt class="py-name">beautiful_soup_tree</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt><tt class="py-op">[</tt><tt class="py-name">last_element_idx</tt><tt class="py-op">+</tt><tt class="py-number">1</tt><tt class="py-op">:</tt><tt class="py-op">]</tt> </tt>
<a name="L139"></a><tt class="py-lineno">139</tt> <tt class="py-line"> </tt>
<a name="L140"></a><tt class="py-lineno">140</tt> <tt class="py-line"> <tt class="py-comment"># Reorganize so that there is one <html> root...</tt> </tt>
<a name="L141"></a><tt class="py-lineno">141</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">html_root</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L142"></a><tt class="py-lineno">142</tt> <tt class="py-line"> <tt class="py-comment"># ... use existing one if possible, ...</tt> </tt>
<a name="L143"></a><tt class="py-lineno">143</tt> <tt class="py-line"> <tt class="py-name">i</tt> <tt class="py-op">=</tt> <tt class="py-name">roots</tt><tt class="py-op">.</tt><tt id="link-47" class="py-name" targets="Method lxml.etree._Element.index()=lxml.etree._Element-class.html#index"><a title="lxml.etree._Element.index" class="py-name" href="#" onclick="return doclink('link-47', 'index', 'link-47');">index</a></tt><tt class="py-op">(</tt><tt class="py-name">html_root</tt><tt class="py-op">)</tt> </tt>
<a name="L144"></a><tt class="py-lineno">144</tt> <tt class="py-line"> <tt class="py-name">html_root</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt> <tt class="py-op">=</tt> <tt class="py-name">roots</tt><tt class="py-op">[</tt><tt class="py-op">:</tt><tt class="py-name">i</tt><tt class="py-op">]</tt> <tt class="py-op">+</tt> <tt class="py-name">html_root</tt><tt class="py-op">.</tt><tt class="py-name">contents</tt> <tt class="py-op">+</tt> <tt class="py-name">roots</tt><tt class="py-op">[</tt><tt class="py-name">i</tt><tt class="py-op">+</tt><tt class="py-number">1</tt><tt class="py-op">:</tt><tt class="py-op">]</tt> </tt>
<a name="L145"></a><tt class="py-lineno">145</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L146"></a><tt class="py-lineno">146</tt> <tt class="py-line"> <tt class="py-comment"># ... otherwise create a new one.</tt> </tt>
<a name="L147"></a><tt class="py-lineno">147</tt> <tt class="py-line"> <tt class="py-name">html_root</tt> <tt class="py-op">=</tt> <tt id="link-48" class="py-name" targets="Class lxml.html.soupparser._PseudoTag=lxml.html.soupparser._PseudoTag-class.html"><a title="lxml.html.soupparser._PseudoTag" class="py-name" href="#" onclick="return doclink('link-48', '_PseudoTag', 'link-48');">_PseudoTag</a></tt><tt class="py-op">(</tt><tt class="py-name">roots</tt><tt class="py-op">)</tt> </tt>
<a name="L148"></a><tt class="py-lineno">148</tt> <tt class="py-line"> </tt>
<a name="L149"></a><tt class="py-lineno">149</tt> <tt class="py-line"> <tt class="py-name">convert_node</tt> <tt class="py-op">=</tt> <tt id="link-49" class="py-name" targets="Function lxml.html.soupparser._init_node_converters()=lxml.html.soupparser-module.html#_init_node_converters"><a title="lxml.html.soupparser._init_node_converters" class="py-name" href="#" onclick="return doclink('link-49', '_init_node_converters', 'link-49');">_init_node_converters</a></tt><tt class="py-op">(</tt><tt id="link-50" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-50', 'makeelement', 'link-9');">makeelement</a></tt><tt class="py-op">)</tt> </tt>
<a name="L150"></a><tt class="py-lineno">150</tt> <tt class="py-line"> </tt>
<a name="L151"></a><tt class="py-lineno">151</tt> <tt class="py-line"> <tt class="py-comment"># Process pre_root</tt> </tt>
<a name="L152"></a><tt class="py-lineno">152</tt> <tt class="py-line"> <tt class="py-name">res_root</tt> <tt class="py-op">=</tt> <tt class="py-name">convert_node</tt><tt class="py-op">(</tt><tt class="py-name">html_root</tt><tt class="py-op">)</tt> </tt>
<a name="L153"></a><tt class="py-lineno">153</tt> <tt class="py-line"> <tt class="py-name">prev</tt> <tt class="py-op">=</tt> <tt class="py-name">res_root</tt> </tt>
<a name="L154"></a><tt class="py-lineno">154</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">e</tt> <tt class="py-keyword">in</tt> <tt class="py-name">reversed</tt><tt class="py-op">(</tt><tt class="py-name">pre_root</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L155"></a><tt class="py-lineno">155</tt> <tt class="py-line"> <tt class="py-name">converted</tt> <tt class="py-op">=</tt> <tt class="py-name">convert_node</tt><tt class="py-op">(</tt><tt class="py-name">e</tt><tt class="py-op">)</tt> </tt>
<a name="L156"></a><tt class="py-lineno">156</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">converted</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L157"></a><tt class="py-lineno">157</tt> <tt class="py-line"> <tt class="py-name">prev</tt><tt class="py-op">.</tt><tt id="link-51" class="py-name" targets="Method lxml.etree._Element.addprevious()=lxml.etree._Element-class.html#addprevious"><a title="lxml.etree._Element.addprevious" class="py-name" href="#" onclick="return doclink('link-51', 'addprevious', 'link-51');">addprevious</a></tt><tt class="py-op">(</tt><tt class="py-name">converted</tt><tt class="py-op">)</tt> </tt>
<a name="L158"></a><tt class="py-lineno">158</tt> <tt class="py-line"> <tt class="py-name">prev</tt> <tt class="py-op">=</tt> <tt class="py-name">converted</tt> </tt>
<a name="L159"></a><tt class="py-lineno">159</tt> <tt class="py-line"> </tt>
<a name="L160"></a><tt class="py-lineno">160</tt> <tt class="py-line"> <tt class="py-comment"># ditto for post_root</tt> </tt>
<a name="L161"></a><tt class="py-lineno">161</tt> <tt class="py-line"> <tt class="py-name">prev</tt> <tt class="py-op">=</tt> <tt class="py-name">res_root</tt> </tt>
<a name="L162"></a><tt class="py-lineno">162</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">e</tt> <tt class="py-keyword">in</tt> <tt class="py-name">post_root</tt><tt class="py-op">:</tt> </tt>
<a name="L163"></a><tt class="py-lineno">163</tt> <tt class="py-line"> <tt class="py-name">converted</tt> <tt class="py-op">=</tt> <tt class="py-name">convert_node</tt><tt class="py-op">(</tt><tt class="py-name">e</tt><tt class="py-op">)</tt> </tt>
<a name="L164"></a><tt class="py-lineno">164</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">converted</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L165"></a><tt class="py-lineno">165</tt> <tt class="py-line"> <tt class="py-name">prev</tt><tt class="py-op">.</tt><tt id="link-52" class="py-name" targets="Method lxml.etree._Element.addnext()=lxml.etree._Element-class.html#addnext"><a title="lxml.etree._Element.addnext" class="py-name" href="#" onclick="return doclink('link-52', 'addnext', 'link-52');">addnext</a></tt><tt class="py-op">(</tt><tt class="py-name">converted</tt><tt class="py-op">)</tt> </tt>
<a name="L166"></a><tt class="py-lineno">166</tt> <tt class="py-line"> <tt class="py-name">prev</tt> <tt class="py-op">=</tt> <tt class="py-name">converted</tt> </tt>
<a name="L167"></a><tt class="py-lineno">167</tt> <tt class="py-line"> </tt>
<a name="L168"></a><tt class="py-lineno">168</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">declaration</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L169"></a><tt class="py-lineno">169</tt> <tt class="py-line"> <tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L170"></a><tt class="py-lineno">170</tt> <tt class="py-line"> <tt class="py-comment"># bs4 provides full Doctype string</tt> </tt>
<a name="L171"></a><tt class="py-lineno">171</tt> <tt class="py-line"> <tt class="py-name">doctype_string</tt> <tt class="py-op">=</tt> <tt class="py-name">declaration</tt><tt class="py-op">.</tt><tt class="py-name">output_ready</tt><tt class="py-op">(</tt><tt class="py-op">)</tt> </tt>
<a name="L172"></a><tt class="py-lineno">172</tt> <tt class="py-line"> <tt class="py-keyword">except</tt> <tt class="py-name">AttributeError</tt><tt class="py-op">:</tt> </tt>
<a name="L173"></a><tt class="py-lineno">173</tt> <tt class="py-line"> <tt class="py-name">doctype_string</tt> <tt class="py-op">=</tt> <tt class="py-name">declaration</tt><tt class="py-op">.</tt><tt class="py-name">string</tt> </tt>
<a name="L174"></a><tt class="py-lineno">174</tt> <tt class="py-line"> </tt>
<a name="L175"></a><tt class="py-lineno">175</tt> <tt class="py-line"> <tt class="py-name">match</tt> <tt class="py-op">=</tt> <tt id="link-53" class="py-name"><a title="lxml.html.soupparser._parse_doctype_declaration" class="py-name" href="#" onclick="return doclink('link-53', '_parse_doctype_declaration', 'link-36');">_parse_doctype_declaration</a></tt><tt class="py-op">(</tt><tt class="py-name">doctype_string</tt><tt class="py-op">)</tt> </tt>
<a name="L176"></a><tt class="py-lineno">176</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-keyword">not</tt> <tt class="py-name">match</tt><tt class="py-op">:</tt> </tt>
<a name="L177"></a><tt class="py-lineno">177</tt> <tt class="py-line"> <tt class="py-comment"># Something is wrong if we end up in here. Since soupparser should</tt> </tt>
<a name="L178"></a><tt class="py-lineno">178</tt> <tt class="py-line"> <tt class="py-comment"># tolerate errors, do not raise Exception, just let it pass.</tt> </tt>
<a name="L179"></a><tt class="py-lineno">179</tt> <tt class="py-line"> <tt class="py-keyword">pass</tt> </tt>
<a name="L180"></a><tt class="py-lineno">180</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L181"></a><tt class="py-lineno">181</tt> <tt class="py-line"> <tt id="link-54" class="py-name" targets="Variable lxml.etree.DTD.external_id=lxml.etree.DTD-class.html#external_id"><a title="lxml.etree.DTD.external_id" class="py-name" href="#" onclick="return doclink('link-54', 'external_id', 'link-54');">external_id</a></tt><tt class="py-op">,</tt> <tt class="py-name">sys_uri</tt> <tt class="py-op">=</tt> <tt class="py-name">match</tt><tt class="py-op">.</tt><tt class="py-name">groups</tt><tt class="py-op">(</tt><tt class="py-op">)</tt> </tt>
<a name="L182"></a><tt class="py-lineno">182</tt> <tt class="py-line"> <tt id="link-55" class="py-name" targets="Variable lxml.etree._ElementTree.docinfo=lxml.etree._ElementTree-class.html#docinfo"><a title="lxml.etree._ElementTree.docinfo" class="py-name" href="#" onclick="return doclink('link-55', 'docinfo', 'link-55');">docinfo</a></tt> <tt class="py-op">=</tt> <tt class="py-name">res_root</tt><tt class="py-op">.</tt><tt id="link-56" class="py-name" targets="Method lxml.etree._Element.getroottree()=lxml.etree._Element-class.html#getroottree"><a title="lxml.etree._Element.getroottree" class="py-name" href="#" onclick="return doclink('link-56', 'getroottree', 'link-56');">getroottree</a></tt><tt class="py-op">(</tt><tt class="py-op">)</tt><tt class="py-op">.</tt><tt id="link-57" class="py-name"><a title="lxml.etree._ElementTree.docinfo" class="py-name" href="#" onclick="return doclink('link-57', 'docinfo', 'link-55');">docinfo</a></tt> </tt>
<a name="L183"></a><tt class="py-lineno">183</tt> <tt class="py-line"> <tt class="py-comment"># strip quotes and update DOCTYPE values (any of None, '', '...')</tt> </tt>
<a name="L184"></a><tt class="py-lineno">184</tt> <tt class="py-line"> <tt id="link-58" class="py-name"><a title="lxml.etree._ElementTree.docinfo" class="py-name" href="#" onclick="return doclink('link-58', 'docinfo', 'link-55');">docinfo</a></tt><tt class="py-op">.</tt><tt id="link-59" class="py-name" targets="Variable lxml.etree.DocInfo.public_id=lxml.etree.DocInfo-class.html#public_id"><a title="lxml.etree.DocInfo.public_id" class="py-name" href="#" onclick="return doclink('link-59', 'public_id', 'link-59');">public_id</a></tt> <tt class="py-op">=</tt> <tt id="link-60" class="py-name"><a title="lxml.etree.DTD.external_id" class="py-name" href="#" onclick="return doclink('link-60', 'external_id', 'link-54');">external_id</a></tt> <tt class="py-keyword">and</tt> <tt id="link-61" class="py-name"><a title="lxml.etree.DTD.external_id" class="py-name" href="#" onclick="return doclink('link-61', 'external_id', 'link-54');">external_id</a></tt><tt class="py-op">[</tt><tt class="py-number">1</tt><tt class="py-op">:</tt><tt class="py-op">-</tt><tt class="py-number">1</tt><tt class="py-op">]</tt> </tt>
<a name="L185"></a><tt class="py-lineno">185</tt> <tt class="py-line"> <tt id="link-62" class="py-name"><a title="lxml.etree._ElementTree.docinfo" class="py-name" href="#" onclick="return doclink('link-62', 'docinfo', 'link-55');">docinfo</a></tt><tt class="py-op">.</tt><tt id="link-63" class="py-name" targets="Variable lxml.etree.DTD.system_url=lxml.etree.DTD-class.html#system_url,Variable lxml.etree.DocInfo.system_url=lxml.etree.DocInfo-class.html#system_url"><a title="lxml.etree.DTD.system_url
lxml.etree.DocInfo.system_url" class="py-name" href="#" onclick="return doclink('link-63', 'system_url', 'link-63');">system_url</a></tt> <tt class="py-op">=</tt> <tt class="py-name">sys_uri</tt> <tt class="py-keyword">and</tt> <tt class="py-name">sys_uri</tt><tt class="py-op">[</tt><tt class="py-number">1</tt><tt class="py-op">:</tt><tt class="py-op">-</tt><tt class="py-number">1</tt><tt class="py-op">]</tt> </tt>
<a name="L186"></a><tt class="py-lineno">186</tt> <tt class="py-line"> </tt>
<a name="L187"></a><tt class="py-lineno">187</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">res_root</tt> </tt>
</div><a name="L188"></a><tt class="py-lineno">188</tt> <tt class="py-line"> </tt>
<a name="_init_node_converters"></a><div id="_init_node_converters-def"><a name="L189"></a><tt class="py-lineno">189</tt> <tt class="py-line"> </tt>
<a name="L190"></a><tt class="py-lineno">190</tt> <a class="py-toggle" href="#" id="_init_node_converters-toggle" onclick="return toggle('_init_node_converters');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#_init_node_converters">_init_node_converters</a><tt class="py-op">(</tt><tt class="py-param">makeelement</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="_init_node_converters-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="_init_node_converters-expanded"><a name="L191"></a><tt class="py-lineno">191</tt> <tt class="py-line"> <tt class="py-name">converters</tt> <tt class="py-op">=</tt> <tt class="py-op">{</tt><tt class="py-op">}</tt> </tt>
<a name="L192"></a><tt class="py-lineno">192</tt> <tt class="py-line"> <tt class="py-name">ordered_node_types</tt> <tt class="py-op">=</tt> <tt class="py-op">[</tt><tt class="py-op">]</tt> </tt>
<a name="L193"></a><tt class="py-lineno">193</tt> <tt class="py-line"> </tt>
<a name="L194"></a><tt class="py-lineno">194</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">converter</tt><tt class="py-op">(</tt><tt class="py-op">*</tt><tt class="py-param">types</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L195"></a><tt class="py-lineno">195</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">add</tt><tt class="py-op">(</tt><tt class="py-param">handler</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L196"></a><tt class="py-lineno">196</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">t</tt> <tt class="py-keyword">in</tt> <tt class="py-name">types</tt><tt class="py-op">:</tt> </tt>
<a name="L197"></a><tt class="py-lineno">197</tt> <tt class="py-line"> <tt class="py-name">converters</tt><tt class="py-op">[</tt><tt class="py-name">t</tt><tt class="py-op">]</tt> <tt class="py-op">=</tt> <tt class="py-name">handler</tt> </tt>
<a name="L198"></a><tt class="py-lineno">198</tt> <tt class="py-line"> <tt class="py-name">ordered_node_types</tt><tt class="py-op">.</tt><tt id="link-64" class="py-name" targets="Method lxml.etree._Element.append()=lxml.etree._Element-class.html#append"><a title="lxml.etree._Element.append" class="py-name" href="#" onclick="return doclink('link-64', 'append', 'link-64');">append</a></tt><tt class="py-op">(</tt><tt class="py-name">t</tt><tt class="py-op">)</tt> </tt>
<a name="L199"></a><tt class="py-lineno">199</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">handler</tt> </tt>
</div><a name="L200"></a><tt class="py-lineno">200</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-65" class="py-name" targets="Method lxml.html.CheckboxValues.add()=lxml.html.CheckboxValues-class.html#add,Method lxml.html.Classes.add()=lxml.html.Classes-class.html#add,Method lxml.html.MultipleSelectOptions.add()=lxml.html.MultipleSelectOptions-class.html#add"><a title="lxml.html.CheckboxValues.add
lxml.html.Classes.add
lxml.html.MultipleSelectOptions.add" class="py-name" href="#" onclick="return doclink('link-65', 'add', 'link-65');">add</a></tt> </tt>
</div><a name="L201"></a><tt class="py-lineno">201</tt> <tt class="py-line"> </tt>
<a name="L202"></a><tt class="py-lineno">202</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">find_best_converter</tt><tt class="py-op">(</tt><tt class="py-param">node</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L203"></a><tt class="py-lineno">203</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">t</tt> <tt class="py-keyword">in</tt> <tt class="py-name">ordered_node_types</tt><tt class="py-op">:</tt> </tt>
<a name="L204"></a><tt class="py-lineno">204</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">isinstance</tt><tt class="py-op">(</tt><tt class="py-name">node</tt><tt class="py-op">,</tt> <tt class="py-name">t</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L205"></a><tt class="py-lineno">205</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">converters</tt><tt class="py-op">[</tt><tt class="py-name">t</tt><tt class="py-op">]</tt> </tt>
<a name="L206"></a><tt class="py-lineno">206</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">None</tt> </tt>
</div><a name="L207"></a><tt class="py-lineno">207</tt> <tt class="py-line"> </tt>
<a name="L208"></a><tt class="py-lineno">208</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">convert_node</tt><tt class="py-op">(</tt><tt class="py-param">bs_node</tt><tt class="py-op">,</tt> <tt class="py-param">parent</tt><tt class="py-op">=</tt><tt class="py-name">None</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L209"></a><tt class="py-lineno">209</tt> <tt class="py-line"> <tt class="py-comment"># duplicated in convert_tag() below</tt> </tt>
<a name="L210"></a><tt class="py-lineno">210</tt> <tt class="py-line"> <tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L211"></a><tt class="py-lineno">211</tt> <tt class="py-line"> <tt class="py-name">handler</tt> <tt class="py-op">=</tt> <tt class="py-name">converters</tt><tt class="py-op">[</tt><tt id="link-66" class="py-name" targets="Variable lxml.etree._LogEntry.type=lxml.etree._LogEntry-class.html#type,Variable lxml.html.InputElement.type=lxml.html.InputElement-class.html#type"><a title="lxml.etree._LogEntry.type
lxml.html.InputElement.type" class="py-name" href="#" onclick="return doclink('link-66', 'type', 'link-66');">type</a></tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">)</tt><tt class="py-op">]</tt> </tt>
<a name="L212"></a><tt class="py-lineno">212</tt> <tt class="py-line"> <tt class="py-keyword">except</tt> <tt class="py-name">KeyError</tt><tt class="py-op">:</tt> </tt>
<a name="L213"></a><tt class="py-lineno">213</tt> <tt class="py-line"> <tt class="py-name">handler</tt> <tt class="py-op">=</tt> <tt class="py-name">converters</tt><tt class="py-op">[</tt><tt id="link-67" class="py-name"><a title="lxml.etree._LogEntry.type
lxml.html.InputElement.type" class="py-name" href="#" onclick="return doclink('link-67', 'type', 'link-66');">type</a></tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">)</tt><tt class="py-op">]</tt> <tt class="py-op">=</tt> <tt class="py-name">find_best_converter</tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">)</tt> </tt>
<a name="L214"></a><tt class="py-lineno">214</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">handler</tt> <tt class="py-keyword">is</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L215"></a><tt class="py-lineno">215</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">None</tt> </tt>
<a name="L216"></a><tt class="py-lineno">216</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">handler</tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">,</tt> <tt class="py-name">parent</tt><tt class="py-op">)</tt> </tt>
</div><a name="L217"></a><tt class="py-lineno">217</tt> <tt class="py-line"> </tt>
<a name="L218"></a><tt class="py-lineno">218</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">map_attrs</tt><tt class="py-op">(</tt><tt class="py-param">bs_attrs</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L219"></a><tt class="py-lineno">219</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">isinstance</tt><tt class="py-op">(</tt><tt class="py-name">bs_attrs</tt><tt class="py-op">,</tt> <tt class="py-name">dict</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> <tt class="py-comment"># bs4</tt> </tt>
<a name="L220"></a><tt class="py-lineno">220</tt> <tt class="py-line"> <tt class="py-name">attribs</tt> <tt class="py-op">=</tt> <tt class="py-op">{</tt><tt class="py-op">}</tt> </tt>
<a name="L221"></a><tt class="py-lineno">221</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">k</tt><tt class="py-op">,</tt> <tt id="link-68" class="py-name" targets="Variable lxml.tests.test_objectify.v=lxml.tests.test_objectify-module.html#v"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-68', 'v', 'link-68');">v</a></tt> <tt class="py-keyword">in</tt> <tt class="py-name">bs_attrs</tt><tt class="py-op">.</tt><tt id="link-69" class="py-name" targets="Method lxml.etree._Attrib.items()=lxml.etree._Attrib-class.html#items,Method lxml.etree._Element.items()=lxml.etree._Element-class.html#items,Method lxml.etree._IDDict.items()=lxml.etree._IDDict-class.html#items"><a title="lxml.etree._Attrib.items
lxml.etree._Element.items
lxml.etree._IDDict.items" class="py-name" href="#" onclick="return doclink('link-69', 'items', 'link-69');">items</a></tt><tt class="py-op">(</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L222"></a><tt class="py-lineno">222</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">isinstance</tt><tt class="py-op">(</tt><tt id="link-70" class="py-name"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-70', 'v', 'link-68');">v</a></tt><tt class="py-op">,</tt> <tt class="py-name">list</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L223"></a><tt class="py-lineno">223</tt> <tt class="py-line"> <tt id="link-71" class="py-name"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-71', 'v', 'link-68');">v</a></tt> <tt class="py-op">=</tt> <tt class="py-string">" "</tt><tt class="py-op">.</tt><tt class="py-name">join</tt><tt class="py-op">(</tt><tt id="link-72" class="py-name"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-72', 'v', 'link-68');">v</a></tt><tt class="py-op">)</tt> </tt>
<a name="L224"></a><tt class="py-lineno">224</tt> <tt class="py-line"> <tt class="py-name">attribs</tt><tt class="py-op">[</tt><tt class="py-name">k</tt><tt class="py-op">]</tt> <tt class="py-op">=</tt> <tt id="link-73" class="py-name" targets="Function lxml.html.soupparser.unescape()=lxml.html.soupparser-module.html#unescape"><a title="lxml.html.soupparser.unescape" class="py-name" href="#" onclick="return doclink('link-73', 'unescape', 'link-73');">unescape</a></tt><tt class="py-op">(</tt><tt id="link-74" class="py-name"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-74', 'v', 'link-68');">v</a></tt><tt class="py-op">)</tt> </tt>
<a name="L225"></a><tt class="py-lineno">225</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L226"></a><tt class="py-lineno">226</tt> <tt class="py-line"> <tt class="py-name">attribs</tt> <tt class="py-op">=</tt> <tt class="py-name">dict</tt><tt class="py-op">(</tt><tt class="py-op">(</tt><tt class="py-name">k</tt><tt class="py-op">,</tt> <tt id="link-75" class="py-name"><a title="lxml.html.soupparser.unescape" class="py-name" href="#" onclick="return doclink('link-75', 'unescape', 'link-73');">unescape</a></tt><tt class="py-op">(</tt><tt id="link-76" class="py-name"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-76', 'v', 'link-68');">v</a></tt><tt class="py-op">)</tt><tt class="py-op">)</tt> <tt class="py-keyword">for</tt> <tt class="py-name">k</tt><tt class="py-op">,</tt> <tt id="link-77" class="py-name"><a title="lxml.tests.test_objectify.v" class="py-name" href="#" onclick="return doclink('link-77', 'v', 'link-68');">v</a></tt> <tt class="py-keyword">in</tt> <tt class="py-name">bs_attrs</tt><tt class="py-op">)</tt> </tt>
<a name="L227"></a><tt class="py-lineno">227</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">attribs</tt> </tt>
</div><a name="L228"></a><tt class="py-lineno">228</tt> <tt class="py-line"> </tt>
<a name="L229"></a><tt class="py-lineno">229</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">append_text</tt><tt class="py-op">(</tt><tt class="py-param">parent</tt><tt class="py-op">,</tt> <tt class="py-param">text</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L230"></a><tt class="py-lineno">230</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">len</tt><tt class="py-op">(</tt><tt class="py-name">parent</tt><tt class="py-op">)</tt> <tt class="py-op">==</tt> <tt class="py-number">0</tt><tt class="py-op">:</tt> </tt>
<a name="L231"></a><tt class="py-lineno">231</tt> <tt class="py-line"> <tt class="py-name">parent</tt><tt class="py-op">.</tt><tt id="link-78" class="py-name" targets="Variable lxml.etree.QName.text=lxml.etree.QName-class.html#text,Variable lxml.etree._Element.text=lxml.etree._Element-class.html#text,Variable lxml.etree._Entity.text=lxml.etree._Entity-class.html#text,Variable lxml.objectify.ObjectifiedElement.text=lxml.objectify.ObjectifiedElement-class.html#text,Variable xml.etree.ElementTree.Element.text=xml.etree.ElementTree.Element-class.html#text"><a title="lxml.etree.QName.text
lxml.etree._Element.text
lxml.etree._Entity.text
lxml.objectify.ObjectifiedElement.text
xml.etree.ElementTree.Element.text" class="py-name" href="#" onclick="return doclink('link-78', 'text', 'link-78');">text</a></tt> <tt class="py-op">=</tt> <tt class="py-op">(</tt><tt class="py-name">parent</tt><tt class="py-op">.</tt><tt id="link-79" class="py-name"><a title="lxml.etree.QName.text
lxml.etree._Element.text
lxml.etree._Entity.text
lxml.objectify.ObjectifiedElement.text
xml.etree.ElementTree.Element.text" class="py-name" href="#" onclick="return doclink('link-79', 'text', 'link-78');">text</a></tt> <tt class="py-keyword">or</tt> <tt class="py-string">''</tt><tt class="py-op">)</tt> <tt class="py-op">+</tt> <tt id="link-80" class="py-name"><a title="lxml.etree.QName.text
lxml.etree._Element.text
lxml.etree._Entity.text
lxml.objectify.ObjectifiedElement.text
xml.etree.ElementTree.Element.text" class="py-name" href="#" onclick="return doclink('link-80', 'text', 'link-78');">text</a></tt> </tt>
<a name="L232"></a><tt class="py-lineno">232</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L233"></a><tt class="py-lineno">233</tt> <tt class="py-line"> <tt class="py-name">parent</tt><tt class="py-op">[</tt><tt class="py-op">-</tt><tt class="py-number">1</tt><tt class="py-op">]</tt><tt class="py-op">.</tt><tt id="link-81" class="py-name" targets="Variable lxml.etree._Element.tail=lxml.etree._Element-class.html#tail,Variable xml.etree.ElementTree.Element.tail=xml.etree.ElementTree.Element-class.html#tail"><a title="lxml.etree._Element.tail
xml.etree.ElementTree.Element.tail" class="py-name" href="#" onclick="return doclink('link-81', 'tail', 'link-81');">tail</a></tt> <tt class="py-op">=</tt> <tt class="py-op">(</tt><tt class="py-name">parent</tt><tt class="py-op">[</tt><tt class="py-op">-</tt><tt class="py-number">1</tt><tt class="py-op">]</tt><tt class="py-op">.</tt><tt id="link-82" class="py-name"><a title="lxml.etree._Element.tail
xml.etree.ElementTree.Element.tail" class="py-name" href="#" onclick="return doclink('link-82', 'tail', 'link-81');">tail</a></tt> <tt class="py-keyword">or</tt> <tt class="py-string">''</tt><tt class="py-op">)</tt> <tt class="py-op">+</tt> <tt id="link-83" class="py-name"><a title="lxml.etree.QName.text
lxml.etree._Element.text
lxml.etree._Entity.text
lxml.objectify.ObjectifiedElement.text
xml.etree.ElementTree.Element.text" class="py-name" href="#" onclick="return doclink('link-83', 'text', 'link-78');">text</a></tt> </tt>
</div><a name="L234"></a><tt class="py-lineno">234</tt> <tt class="py-line"> </tt>
<a name="L235"></a><tt class="py-lineno">235</tt> <tt class="py-line"> <tt class="py-comment"># converters are tried in order of their definition</tt> </tt>
<a name="L236"></a><tt class="py-lineno">236</tt> <tt class="py-line"> </tt>
<a name="L237"></a><tt class="py-lineno">237</tt> <tt class="py-line"> <tt class="py-decorator">@</tt><tt class="py-decorator">converter</tt><tt class="py-op">(</tt><tt class="py-name">Tag</tt><tt class="py-op">,</tt> <tt id="link-84" class="py-name"><a title="lxml.html.soupparser._PseudoTag" class="py-name" href="#" onclick="return doclink('link-84', '_PseudoTag', 'link-48');">_PseudoTag</a></tt><tt class="py-op">)</tt> </tt>
<a name="L238"></a><tt class="py-lineno">238</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">convert_tag</tt><tt class="py-op">(</tt><tt class="py-param">bs_node</tt><tt class="py-op">,</tt> <tt class="py-param">parent</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L239"></a><tt class="py-lineno">239</tt> <tt class="py-line"> <tt class="py-name">attrs</tt> <tt class="py-op">=</tt> <tt class="py-name">bs_node</tt><tt class="py-op">.</tt><tt class="py-name">attrs</tt> </tt>
<a name="L240"></a><tt class="py-lineno">240</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">parent</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L241"></a><tt class="py-lineno">241</tt> <tt class="py-line"> <tt class="py-name">attribs</tt> <tt class="py-op">=</tt> <tt class="py-name">map_attrs</tt><tt class="py-op">(</tt><tt class="py-name">attrs</tt><tt class="py-op">)</tt> <tt class="py-keyword">if</tt> <tt class="py-name">attrs</tt> <tt class="py-keyword">else</tt> <tt class="py-name">None</tt> </tt>
<a name="L242"></a><tt class="py-lineno">242</tt> <tt class="py-line"> <tt class="py-name">res</tt> <tt class="py-op">=</tt> <tt id="link-85" class="py-name"><a title="lxml.etree
lxml.sax.ElementTreeContentHandler.etree
lxml.tests.test_elementtree.CElementTreeTestCase.etree
lxml.tests.test_elementtree._ETreeTestCaseBase.etree
lxml.tests.test_elementtree._XMLPullParserTest.etree
lxml.tests.test_io._IOTestCaseBase.etree" class="py-name" href="#" onclick="return doclink('link-85', 'etree', 'link-1');">etree</a></tt><tt class="py-op">.</tt><tt id="link-86" class="py-name" targets="Function lxml.etree.SubElement()=lxml.etree-module.html#SubElement"><a title="lxml.etree.SubElement" class="py-name" href="#" onclick="return doclink('link-86', 'SubElement', 'link-86');">SubElement</a></tt><tt class="py-op">(</tt><tt class="py-name">parent</tt><tt class="py-op">,</tt> <tt class="py-name">bs_node</tt><tt class="py-op">.</tt><tt id="link-87" class="py-name"><a title="lxml.etree.DTD.name
lxml.etree._Entity.name
lxml.html.InputMixin.name
lxml.objectify.PyType.name" class="py-name" href="#" onclick="return doclink('link-87', 'name', 'link-37');">name</a></tt><tt class="py-op">,</tt> <tt id="link-88" class="py-name" targets="Variable lxml.etree._Element.attrib=lxml.etree._Element-class.html#attrib,Variable lxml.etree._ProcessingInstruction.attrib=lxml.etree._ProcessingInstruction-class.html#attrib,Function lxml.tests.selftest2.attrib()=lxml.tests.selftest2-module.html#attrib,Variable xml.etree.ElementTree.Element.attrib=xml.etree.ElementTree.Element-class.html#attrib"><a title="lxml.etree._Element.attrib
lxml.etree._ProcessingInstruction.attrib
lxml.tests.selftest2.attrib
xml.etree.ElementTree.Element.attrib" class="py-name" href="#" onclick="return doclink('link-88', 'attrib', 'link-88');">attrib</a></tt><tt class="py-op">=</tt><tt class="py-name">attribs</tt><tt class="py-op">)</tt> </tt>
<a name="L243"></a><tt class="py-lineno">243</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L244"></a><tt class="py-lineno">244</tt> <tt class="py-line"> <tt class="py-name">attribs</tt> <tt class="py-op">=</tt> <tt class="py-name">map_attrs</tt><tt class="py-op">(</tt><tt class="py-name">attrs</tt><tt class="py-op">)</tt> <tt class="py-keyword">if</tt> <tt class="py-name">attrs</tt> <tt class="py-keyword">else</tt> <tt class="py-op">{</tt><tt class="py-op">}</tt> </tt>
<a name="L245"></a><tt class="py-lineno">245</tt> <tt class="py-line"> <tt class="py-name">res</tt> <tt class="py-op">=</tt> <tt id="link-89" class="py-name"><a title="lxml.etree._Element.makeelement
lxml.etree.iterparse.makeelement
lxml.tests.selftest2.makeelement" class="py-name" href="#" onclick="return doclink('link-89', 'makeelement', 'link-9');">makeelement</a></tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">.</tt><tt id="link-90" class="py-name"><a title="lxml.etree.DTD.name
lxml.etree._Entity.name
lxml.html.InputMixin.name
lxml.objectify.PyType.name" class="py-name" href="#" onclick="return doclink('link-90', 'name', 'link-37');">name</a></tt><tt class="py-op">,</tt> <tt id="link-91" class="py-name"><a title="lxml.etree._Element.attrib
lxml.etree._ProcessingInstruction.attrib
lxml.tests.selftest2.attrib
xml.etree.ElementTree.Element.attrib" class="py-name" href="#" onclick="return doclink('link-91', 'attrib', 'link-88');">attrib</a></tt><tt class="py-op">=</tt><tt class="py-name">attribs</tt><tt class="py-op">)</tt> </tt>
<a name="L246"></a><tt class="py-lineno">246</tt> <tt class="py-line"> </tt>
<a name="L247"></a><tt class="py-lineno">247</tt> <tt class="py-line"> <tt class="py-keyword">for</tt> <tt class="py-name">child</tt> <tt class="py-keyword">in</tt> <tt class="py-name">bs_node</tt><tt class="py-op">:</tt> </tt>
<a name="L248"></a><tt class="py-lineno">248</tt> <tt class="py-line"> <tt class="py-comment"># avoid double recursion by inlining convert_node(), see above</tt> </tt>
<a name="L249"></a><tt class="py-lineno">249</tt> <tt class="py-line"> <tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L250"></a><tt class="py-lineno">250</tt> <tt class="py-line"> <tt class="py-name">handler</tt> <tt class="py-op">=</tt> <tt class="py-name">converters</tt><tt class="py-op">[</tt><tt id="link-92" class="py-name"><a title="lxml.etree._LogEntry.type
lxml.html.InputElement.type" class="py-name" href="#" onclick="return doclink('link-92', 'type', 'link-66');">type</a></tt><tt class="py-op">(</tt><tt class="py-name">child</tt><tt class="py-op">)</tt><tt class="py-op">]</tt> </tt>
<a name="L251"></a><tt class="py-lineno">251</tt> <tt class="py-line"> <tt class="py-keyword">except</tt> <tt class="py-name">KeyError</tt><tt class="py-op">:</tt> </tt>
<a name="L252"></a><tt class="py-lineno">252</tt> <tt class="py-line"> <tt class="py-keyword">pass</tt> </tt>
<a name="L253"></a><tt class="py-lineno">253</tt> <tt class="py-line"> <tt class="py-keyword">else</tt><tt class="py-op">:</tt> </tt>
<a name="L254"></a><tt class="py-lineno">254</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">handler</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L255"></a><tt class="py-lineno">255</tt> <tt class="py-line"> <tt class="py-name">handler</tt><tt class="py-op">(</tt><tt class="py-name">child</tt><tt class="py-op">,</tt> <tt class="py-name">res</tt><tt class="py-op">)</tt> </tt>
<a name="L256"></a><tt class="py-lineno">256</tt> <tt class="py-line"> <tt class="py-keyword">continue</tt> </tt>
<a name="L257"></a><tt class="py-lineno">257</tt> <tt class="py-line"> <tt class="py-name">convert_node</tt><tt class="py-op">(</tt><tt class="py-name">child</tt><tt class="py-op">,</tt> <tt class="py-name">res</tt><tt class="py-op">)</tt> </tt>
<a name="L258"></a><tt class="py-lineno">258</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">res</tt> </tt>
</div><a name="L259"></a><tt class="py-lineno">259</tt> <tt class="py-line"> </tt>
<a name="L260"></a><tt class="py-lineno">260</tt> <tt class="py-line"> <tt class="py-decorator">@</tt><tt class="py-decorator">converter</tt><tt class="py-op">(</tt><tt id="link-93" class="py-name"><a title="lxml.etree.Comment" class="py-name" href="#" onclick="return doclink('link-93', 'Comment', 'link-3');">Comment</a></tt><tt class="py-op">)</tt> </tt>
<a name="L261"></a><tt class="py-lineno">261</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">convert_comment</tt><tt class="py-op">(</tt><tt class="py-param">bs_node</tt><tt class="py-op">,</tt> <tt class="py-param">parent</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L262"></a><tt class="py-lineno">262</tt> <tt class="py-line"> <tt class="py-name">res</tt> <tt class="py-op">=</tt> <tt id="link-94" class="py-name"><a title="lxml.html
lxml.html.diff.href_token.html
lxml.html.diff.tag_token.html
lxml.html.diff.token.html" class="py-name" href="#" onclick="return doclink('link-94', 'html', 'link-2');">html</a></tt><tt class="py-op">.</tt><tt id="link-95" class="py-name" targets="Class lxml.html.HtmlComment=lxml.html.HtmlComment-class.html"><a title="lxml.html.HtmlComment" class="py-name" href="#" onclick="return doclink('link-95', 'HtmlComment', 'link-95');">HtmlComment</a></tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">)</tt> </tt>
<a name="L263"></a><tt class="py-lineno">263</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">parent</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L264"></a><tt class="py-lineno">264</tt> <tt class="py-line"> <tt class="py-name">parent</tt><tt class="py-op">.</tt><tt id="link-96" class="py-name"><a title="lxml.etree._Element.append" class="py-name" href="#" onclick="return doclink('link-96', 'append', 'link-64');">append</a></tt><tt class="py-op">(</tt><tt class="py-name">res</tt><tt class="py-op">)</tt> </tt>
<a name="L265"></a><tt class="py-lineno">265</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">res</tt> </tt>
</div><a name="L266"></a><tt class="py-lineno">266</tt> <tt class="py-line"> </tt>
<a name="L267"></a><tt class="py-lineno">267</tt> <tt class="py-line"> <tt class="py-decorator">@</tt><tt class="py-decorator">converter</tt><tt class="py-op">(</tt><tt class="py-name">ProcessingInstruction</tt><tt class="py-op">)</tt> </tt>
<a name="L268"></a><tt class="py-lineno">268</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">convert_pi</tt><tt class="py-op">(</tt><tt class="py-param">bs_node</tt><tt class="py-op">,</tt> <tt class="py-param">parent</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L269"></a><tt class="py-lineno">269</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">bs_node</tt><tt class="py-op">.</tt><tt class="py-name">endswith</tt><tt class="py-op">(</tt><tt class="py-string">'?'</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L270"></a><tt class="py-lineno">270</tt> <tt class="py-line"> <tt class="py-comment"># The PI is of XML style (<?as df?>) but BeautifulSoup</tt> </tt>
<a name="L271"></a><tt class="py-lineno">271</tt> <tt class="py-line"> <tt class="py-comment"># interpreted it as being SGML style (<?as df>). Fix.</tt> </tt>
<a name="L272"></a><tt class="py-lineno">272</tt> <tt class="py-line"> <tt class="py-name">bs_node</tt> <tt class="py-op">=</tt> <tt class="py-name">bs_node</tt><tt class="py-op">[</tt><tt class="py-op">:</tt><tt class="py-op">-</tt><tt class="py-number">1</tt><tt class="py-op">]</tt> </tt>
<a name="L273"></a><tt class="py-lineno">273</tt> <tt class="py-line"> <tt class="py-name">res</tt> <tt class="py-op">=</tt> <tt id="link-97" class="py-name"><a title="lxml.etree
lxml.sax.ElementTreeContentHandler.etree
lxml.tests.test_elementtree.CElementTreeTestCase.etree
lxml.tests.test_elementtree._ETreeTestCaseBase.etree
lxml.tests.test_elementtree._XMLPullParserTest.etree
lxml.tests.test_io._IOTestCaseBase.etree" class="py-name" href="#" onclick="return doclink('link-97', 'etree', 'link-1');">etree</a></tt><tt class="py-op">.</tt><tt class="py-name">ProcessingInstruction</tt><tt class="py-op">(</tt><tt class="py-op">*</tt><tt class="py-name">bs_node</tt><tt class="py-op">.</tt><tt class="py-name">split</tt><tt class="py-op">(</tt><tt class="py-string">' '</tt><tt class="py-op">,</tt> <tt class="py-number">1</tt><tt class="py-op">)</tt><tt class="py-op">)</tt> </tt>
<a name="L274"></a><tt class="py-lineno">274</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">parent</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L275"></a><tt class="py-lineno">275</tt> <tt class="py-line"> <tt class="py-name">parent</tt><tt class="py-op">.</tt><tt id="link-98" class="py-name"><a title="lxml.etree._Element.append" class="py-name" href="#" onclick="return doclink('link-98', 'append', 'link-64');">append</a></tt><tt class="py-op">(</tt><tt class="py-name">res</tt><tt class="py-op">)</tt> </tt>
<a name="L276"></a><tt class="py-lineno">276</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">res</tt> </tt>
</div><a name="L277"></a><tt class="py-lineno">277</tt> <tt class="py-line"> </tt>
<a name="L278"></a><tt class="py-lineno">278</tt> <tt class="py-line"> <tt class="py-decorator">@</tt><tt class="py-decorator">converter</tt><tt class="py-op">(</tt><tt class="py-name">NavigableString</tt><tt class="py-op">)</tt> </tt>
<a name="L279"></a><tt class="py-lineno">279</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">convert_text</tt><tt class="py-op">(</tt><tt class="py-param">bs_node</tt><tt class="py-op">,</tt> <tt class="py-param">parent</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L280"></a><tt class="py-lineno">280</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-name">parent</tt> <tt class="py-keyword">is</tt> <tt class="py-keyword">not</tt> <tt class="py-name">None</tt><tt class="py-op">:</tt> </tt>
<a name="L281"></a><tt class="py-lineno">281</tt> <tt class="py-line"> <tt class="py-name">append_text</tt><tt class="py-op">(</tt><tt class="py-name">parent</tt><tt class="py-op">,</tt> <tt id="link-99" class="py-name"><a title="lxml.html.soupparser.unescape" class="py-name" href="#" onclick="return doclink('link-99', 'unescape', 'link-73');">unescape</a></tt><tt class="py-op">(</tt><tt class="py-name">bs_node</tt><tt class="py-op">)</tt><tt class="py-op">)</tt> </tt>
<a name="L282"></a><tt class="py-lineno">282</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">None</tt> </tt>
</div><a name="L283"></a><tt class="py-lineno">283</tt> <tt class="py-line"> </tt>
<a name="L284"></a><tt class="py-lineno">284</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">convert_node</tt> </tt>
</div><a name="L285"></a><tt class="py-lineno">285</tt> <tt class="py-line"> </tt>
<a name="L286"></a><tt class="py-lineno">286</tt> <tt class="py-line"> </tt>
<a name="L287"></a><tt class="py-lineno">287</tt> <tt class="py-line"><tt class="py-comment"># copied from ET's ElementSoup</tt> </tt>
<a name="L288"></a><tt class="py-lineno">288</tt> <tt class="py-line"> </tt>
<a name="L289"></a><tt class="py-lineno">289</tt> <tt class="py-line"><tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L290"></a><tt class="py-lineno">290</tt> <tt class="py-line"> <tt class="py-keyword">from</tt> <tt id="link-100" class="py-name"><a title="lxml.html
lxml.html.diff.href_token.html
lxml.html.diff.tag_token.html
lxml.html.diff.token.html" class="py-name" href="#" onclick="return doclink('link-100', 'html', 'link-2');">html</a></tt><tt class="py-op">.</tt><tt id="link-101" class="py-name" targets="Method lxml.etree.DTD.entities()=lxml.etree.DTD-class.html#entities"><a title="lxml.etree.DTD.entities" class="py-name" href="#" onclick="return doclink('link-101', 'entities', 'link-101');">entities</a></tt> <tt class="py-keyword">import</tt> <tt class="py-name">name2codepoint</tt> <tt class="py-comment"># Python 3</tt> </tt>
<a name="L291"></a><tt class="py-lineno">291</tt> <tt class="py-line"><tt class="py-keyword">except</tt> <tt class="py-name">ImportError</tt><tt class="py-op">:</tt> </tt>
<a name="L292"></a><tt class="py-lineno">292</tt> <tt class="py-line"> <tt class="py-keyword">from</tt> <tt class="py-name">htmlentitydefs</tt> <tt class="py-keyword">import</tt> <tt class="py-name">name2codepoint</tt> </tt>
<a name="L293"></a><tt class="py-lineno">293</tt> <tt class="py-line"> </tt>
<a name="L294"></a><tt class="py-lineno">294</tt> <tt class="py-line"> </tt>
<a name="L295"></a><tt class="py-lineno">295</tt> <tt class="py-line"><tt id="link-102" class="py-name" targets="Function lxml.html.soupparser.handle_entities()=lxml.html.soupparser-module.html#handle_entities"><a title="lxml.html.soupparser.handle_entities" class="py-name" href="#" onclick="return doclink('link-102', 'handle_entities', 'link-102');">handle_entities</a></tt> <tt class="py-op">=</tt> <tt class="py-name">re</tt><tt class="py-op">.</tt><tt class="py-name">compile</tt><tt class="py-op">(</tt><tt class="py-string">r"&(\w+);"</tt><tt class="py-op">)</tt><tt class="py-op">.</tt><tt class="py-name">sub</tt> </tt>
<a name="L296"></a><tt class="py-lineno">296</tt> <tt class="py-line"> </tt>
<a name="L297"></a><tt class="py-lineno">297</tt> <tt class="py-line"> </tt>
<a name="L298"></a><tt class="py-lineno">298</tt> <tt class="py-line"><tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L299"></a><tt class="py-lineno">299</tt> <tt class="py-line"> <tt id="link-103" class="py-name" targets="Function lxml.html.soupparser.unichr()=lxml.html.soupparser-module.html#unichr"><a title="lxml.html.soupparser.unichr" class="py-name" href="#" onclick="return doclink('link-103', 'unichr', 'link-103');">unichr</a></tt> </tt>
<a name="L300"></a><tt class="py-lineno">300</tt> <tt class="py-line"><tt class="py-keyword">except</tt> <tt class="py-name">NameError</tt><tt class="py-op">:</tt> </tt>
<a name="L301"></a><tt class="py-lineno">301</tt> <tt class="py-line"> <tt class="py-comment"># Python 3</tt> </tt>
<a name="L302"></a><tt class="py-lineno">302</tt> <tt class="py-line"> <tt id="link-104" class="py-name"><a title="lxml.html.soupparser.unichr" class="py-name" href="#" onclick="return doclink('link-104', 'unichr', 'link-103');">unichr</a></tt> <tt class="py-op">=</tt> <tt class="py-name">chr</tt> </tt>
<a name="unescape"></a><div id="unescape-def"><a name="L303"></a><tt class="py-lineno">303</tt> <tt class="py-line"> </tt>
<a name="L304"></a><tt class="py-lineno">304</tt> <tt class="py-line"> </tt>
<a name="L305"></a><tt class="py-lineno">305</tt> <a class="py-toggle" href="#" id="unescape-toggle" onclick="return toggle('unescape');">-</a><tt class="py-line"><tt class="py-keyword">def</tt> <a class="py-def-name" href="lxml.html.soupparser-module.html#unescape">unescape</a><tt class="py-op">(</tt><tt class="py-param">string</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
</div><div id="unescape-collapsed" style="display:none;" pad="+++" indent="++++"></div><div id="unescape-expanded"><a name="L306"></a><tt class="py-lineno">306</tt> <tt class="py-line"> <tt class="py-keyword">if</tt> <tt class="py-keyword">not</tt> <tt class="py-name">string</tt><tt class="py-op">:</tt> </tt>
<a name="L307"></a><tt class="py-lineno">307</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-string">''</tt> </tt>
<a name="L308"></a><tt class="py-lineno">308</tt> <tt class="py-line"> <tt class="py-comment"># work around oddities in BeautifulSoup's entity handling</tt> </tt>
<a name="L309"></a><tt class="py-lineno">309</tt> <tt class="py-line"> <tt class="py-keyword">def</tt> <tt class="py-def-name">unescape_entity</tt><tt class="py-op">(</tt><tt class="py-param">m</tt><tt class="py-op">)</tt><tt class="py-op">:</tt> </tt>
<a name="L310"></a><tt class="py-lineno">310</tt> <tt class="py-line"> <tt class="py-keyword">try</tt><tt class="py-op">:</tt> </tt>
<a name="L311"></a><tt class="py-lineno">311</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-105" class="py-name"><a title="lxml.html.soupparser.unichr" class="py-name" href="#" onclick="return doclink('link-105', 'unichr', 'link-103');">unichr</a></tt><tt class="py-op">(</tt><tt class="py-name">name2codepoint</tt><tt class="py-op">[</tt><tt class="py-name">m</tt><tt class="py-op">.</tt><tt class="py-name">group</tt><tt class="py-op">(</tt><tt class="py-number">1</tt><tt class="py-op">)</tt><tt class="py-op">]</tt><tt class="py-op">)</tt> </tt>
<a name="L312"></a><tt class="py-lineno">312</tt> <tt class="py-line"> <tt class="py-keyword">except</tt> <tt class="py-name">KeyError</tt><tt class="py-op">:</tt> </tt>
<a name="L313"></a><tt class="py-lineno">313</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt class="py-name">m</tt><tt class="py-op">.</tt><tt class="py-name">group</tt><tt class="py-op">(</tt><tt class="py-number">0</tt><tt class="py-op">)</tt> <tt class="py-comment"># use as is</tt> </tt>
</div><a name="L314"></a><tt class="py-lineno">314</tt> <tt class="py-line"> <tt class="py-keyword">return</tt> <tt id="link-106" class="py-name"><a title="lxml.html.soupparser.handle_entities" class="py-name" href="#" onclick="return doclink('link-106', 'handle_entities', 'link-102');">handle_entities</a></tt><tt class="py-op">(</tt><tt class="py-name">unescape_entity</tt><tt class="py-op">,</tt> <tt class="py-name">string</tt><tt class="py-op">)</tt> </tt>
</div><a name="L315"></a><tt class="py-lineno">315</tt> <tt class="py-line"> </tt><script type="text/javascript">
<!--
expandto(location.href);
// -->
</script>
</pre>
<br />
<!-- ==================== NAVIGATION BAR ==================== -->
<table class="navbar" border="0" width="100%" cellpadding="0"
bgcolor="#a0c0ff" cellspacing="0">
<tr valign="middle">
<!-- Home link -->
<th> <a
href="lxml-module.html">Home</a> </th>
<!-- Tree link -->
<th> <a
href="module-tree.html">Trees</a> </th>
<!-- Index link -->
<th> <a
href="identifier-index.html">Indices</a> </th>
<!-- Help link -->
<th> <a
href="help.html">Help</a> </th>
<!-- Project homepage -->
<th class="navbar" align="right" width="100%">
<table border="0" cellpadding="0" cellspacing="0">
<tr><th class="navbar" align="center"
><a class="navbar" target="_top" href="/">lxml API</a></th>
</tr></table></th>
</tr>
</table>
<table border="0" cellpadding="0" cellspacing="0" width="100%%">
<tr>
<td align="left" class="footer">
Generated by Epydoc 3.0.1
on Wed Jun 27 16:05:05 2018
</td>
<td align="right" class="footer">
<a target="mainFrame" href="http://epydoc.sourceforge.net"
>http://epydoc.sourceforge.net</a>
</td>
</tr>
</table>
<script type="text/javascript">
<!--
// Private objects are initially displayed (because if
// javascript is turned off then we want them to be
// visible); but by default, we want to hide them. So hide
// them unless we have a cookie that says to show them.
checkCookie();
// -->
</script>
</body>
</html>