gronod/headphones
1
0
Fork 1
mirror of https://github.com/rembo10/headphones.git synced 2026-05-02 17:59:28 +01:00
Code Issues Packages Projects Releases Wiki Activity

Include html5lib for BeautifulSoup

Browse Source 
BeautifulSoup needs lxml or html5, have included html5lib.

Also latest BeautifulSoup 4.1.3
This commit is contained in:
Ade
2012-09-06 10:47:07 +12:00
parent 7aac60cecd
commit b8c5782765
136 changed files with 87265 additions and 2428 deletions
Download Patch File Download Diff File Expand all files Collapse all files

261
html5lib/tests/testdata/tree-construction/tricky01.dat vendored Normal file
View File

@@ -0,0 +1,261 @@
#data
<b><p>Bold </b> Not bold</p>
Also not bold.
#errors
#document
| <html>
| <head>
| <body>
| <b>
| <p>
| <b>
| "Bold "
| " Not bold"
| "
Also not bold."
#data
<html>
<font color=red><i>Italic and Red<p>Italic and Red </font> Just italic.</p> Italic only.</i> Plain
<p>I should not be red. <font color=red>Red. <i>Italic and red.</p>
<p>Italic and red. </i> Red.</font> I should not be red.</p>
<b>Bold <i>Bold and italic</b> Only Italic </i> Plain
#errors
#document
| <html>
| <head>
| <body>
| <font>
| color="red"
| <i>
| "Italic and Red"
| <i>
| <p>
| <font>
| color="red"
| "Italic and Red "
| " Just italic."
| " Italic only."
| " Plain
"
| <p>
| "I should not be red. "
| <font>
| color="red"
| "Red. "
| <i>
| "Italic and red."
| <font>
| color="red"
| <i>
| "
"
| <p>
| <font>
| color="red"
| <i>
| "Italic and red. "
| " Red."
| " I should not be red."
| "
"
| <b>
| "Bold "
| <i>
| "Bold and italic"
| <i>
| " Only Italic "
| " Plain"
#data
<html><body>
<p><font size="7">First paragraph.</p>
<p>Second paragraph.</p></font>
<b><p><i>Bold and Italic</b> Italic</p>
#errors
#document
| <html>
| <head>
| <body>
| "
"
| <p>
| <font>
| size="7"
| "First paragraph."
| <font>
| size="7"
| "
"
| <p>
| "Second paragraph."
| "
"
| <b>
| <p>
| <b>
| <i>
| "Bold and Italic"
| <i>
| " Italic"
#data
<html>
<dl>
<dt><b>Boo
<dd>Goo?
</dl>
</html>
#errors
#document
| <html>
| <head>
| <body>
| <dl>
| "
"
| <dt>
| <b>
| "Boo
"
| <dd>
| <b>
| "Goo?
"
| <b>
| "
"
#data
<html><body>
<label><a><div>Hello<div>World</div></a></label>
</body></html>
#errors
#document
| <html>
| <head>
| <body>
| "
"
| <label>
| <a>
| <div>
| <a>
| "Hello"
| <div>
| "World"
| "
"
#data
<table><center> <font>a</center> <img> <tr><td> </td> </tr> </table>
#errors
#document
| <html>
| <head>
| <body>
| <center>
| " "
| <font>
| "a"
| <font>
| <img>
| " "
| <table>
| " "
| <tbody>
| <tr>
| <td>
| " "
| " "
| " "
#data
<table><tr><p><a><p>You should see this text.
#errors
#document
| <html>
| <head>
| <body>
| <p>
| <a>
| <p>
| <a>
| "You should see this text."
| <table>
| <tbody>
| <tr>
#data
<TABLE>
<TR>
<CENTER><CENTER><TD></TD></TR><TR>
<FONT>
<TABLE><tr></tr></TABLE>
</P>
<a></font><font></a>
This page contains an insanely badly-nested tag sequence.
#errors
#document
| <html>
| <head>
| <body>
| <center>
| <center>
| <font>
| "
"
| <table>
| "
"
| <tbody>
| <tr>
| "
"
| <td>
| <tr>
| "
"
| <table>
| <tbody>
| <tr>
| <font>
| "
"
| <p>
| "
"
| <a>
| <a>
| <font>
| <font>
| "
This page contains an insanely badly-nested tag sequence."
#data
<html>
<body>
<b><nobr><div>This text is in a div inside a nobr</nobr>More text that should not be in the nobr, i.e., the
nobr should have closed the div inside it implicitly. </b><pre>A pre tag outside everything else.</pre>
</body>
</html>
#errors
#document
| <html>
| <head>
| <body>
| "
"
| <b>
| <nobr>
| <div>
| <b>
| <nobr>
| "This text is in a div inside a nobr"
| "More text that should not be in the nobr, i.e., the
nobr should have closed the div inside it implicitly. "
| <pre>
| "A pre tag outside everything else."
| "
"