copyleftie
/
gajim-plural
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

[sgala and I] better support of sizes in XHTML, simplify rst generator

This commit is contained in:
Yann Leboulanger 2007-05-29 20:22:41 +00:00
parent fc59489bdf
commit 96e6b34cd7
2 changed files with 321 additions and 172 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

92
src/common/rst_xhtml_generator.py
View File

@ -21,39 +21,70 @@ try:
from docutils import nodes,utils
from docutils.parsers.rst.roles import set_classes
except:
print "Requires docutils 0.4 for set_classes to be available"
def create_xhtml(text):
return None
else:
def jep_reference_role(role, rawtext, text, lineno, inliner,
options={}, content=[]):
'''Role to make handy references to Jabber Enhancement Proposals (JEP).
def pos_int_validator(text):
"""Validates that text can be evaluated as a positive integer."""
result = int(text)
if result < 0:
raise ValueError("Error: value '%(text)s' "
"must be a positive integer")
return result
Use as :JEP:`71` (or jep, or jep-reference).
Modeled after the sample in docutils documentation.
def generate_uri_role( role_name, aliases,
anchor_text, base_url,
interpret_url, validator):
'''Creates and register a uri based "interpreted role".
Those are similar to the RFC, and PEP ones, and take
role_name:
name that will be registered
aliases:
list of alternate names
anchor_text:
text that will be used, together with the role
base_url:
base url for the link
interpret_url:
this, modulo the validated text, will be added to it
validator:
should return the validated text, or raise ValueError
'''
def uri_reference_role(role, rawtext, text, lineno, inliner,
options={}, content=[]):
try:
valid_text = validator(text)
except ValueError, e:
msg = inliner.reporter.error( e.message % dict(text=text), line=lineno)
prb = inliner.problematic(rawtext, rawtext, msg)
return [prb], [msg]
ref = base_url + interpret_url % valid_text
set_classes(options)
node = nodes.reference(rawtext, anchor_text + utils.unescape(text), refuri=ref,
**options)
return [node], []
jep_base_url = 'http://www.jabber.org/jeps/'
jep_url = 'jep-%04d.html'
try:
jepnum = int(text)
if jepnum <= 0:
raise ValueError
except ValueError:
msg = inliner.reporter.error(
'JEP number must be a number greater than or equal to 1; '
'"%s" is invalid.' % text, line=lineno)
prb = inliner.problematic(rawtext, rawtext, msg)
return [prb], [msg]
ref = jep_base_url + jep_url % jepnum
set_classes(options)
node = nodes.reference(rawtext, 'JEP ' + utils.unescape(text), refuri=ref,
**options)
return [node], []
uri_reference_role.__doc__ = """Role to make handy references to URIs.
roles.register_canonical_role('jep-reference', jep_reference_role)
from docutils.parsers.rst.languages.en import roles
roles['jep-reference'] = 'jep-reference'
roles['jep'] = 'jep-reference'
Use as :%(role_name)s:`71` (or any of %(aliases)s).
It will use %(base_url)s+%(interpret_url)s
validator should throw a ValueError, containing optionally
a %%(text)s format, if the interpreted text is not valid.
""" % locals()
roles.register_canonical_role(role_name, uri_reference_role)
from docutils.parsers.rst.languages import en
en.roles[role_name] = role_name
for alias in aliases:
en.roles[alias] = role_name
generate_uri_role('xep-reference', ('jep', 'xep'),
'XEP #', 'http://www.xmpp.org/extensions/', 'xep-%04d.html',
pos_int_validator)
generate_uri_role('gajim-ticket-reference', ('ticket','gtrack'),
'Gajim Ticket #', 'http://trac.gajim.org/ticket/', '%d',
pos_int_validator)
class HTMLGenerator:
'''Really simple HTMLGenerator starting from publish_parts.
@ -108,7 +139,7 @@ else:
if __name__ == '__main__':
print Generator.create_xhtml('''
print "test 1\n", Generator.create_xhtml("""
test::
>>> print 1
@ -118,9 +149,10 @@ test::
this `` should trigger`` should trigger the &nbsp; problem.
''')
print Generator.create_xhtml('''
""")
print "test 2\n", Generator.create_xhtml("""
*test1
test2_
''')
""")
print "test 3\n", Generator.create_xhtml(""":ticket:`316` implements :xep:`71`""")

401
src/htmltextview.py
View File

@ -40,9 +40,9 @@ import time
import urllib2
import operator
if __name__ == '__main__':
from common import i18n
from common import gajim
#from common import i18n
import tooltips
@ -56,13 +56,13 @@ allwhitespace_rx = re.compile('^\\s*$')
display_resolution = 0.3514598*(gtk.gdk.screen_height() /
float(gtk.gdk.screen_height_mm()))
#embryo of CSS classes
# embryo of CSS classes
classes = {
#'system-message':';display: none',
'problematic':';color: red',
}
#styles for elemens
# styles for elements
element_styles = {
'u' : ';text-decoration: underline',
'em' : ';font-style: oblique',
@ -83,9 +83,6 @@ element_styles['tt'] = element_styles['kbd']
element_styles['i'] = element_styles['em']
element_styles['b'] = element_styles['strong']
class_styles = {
}
'''
==========
JEP-0071
@ -182,40 +179,43 @@ for name in BLOCK_HEAD:
def build_patterns(view, config, interface):
#extra, rst does not mark _underline_ or /it/ up
#actually <b>, <i> or <u> are not in the JEP-0071, but are seen in the wild
# extra, rst does not mark _underline_ or /it/ up
# actually <b>, <i> or <u> are not in the JEP-0071, but are seen in the wild
basic_pattern = r'(?<!\w|\<|/|:)' r'/[^\s/]' r'([^/]*[^\s/])?' r'/(?!\w|/|:)|'\
r'(?<!\w)' r'_[^\s_]' r'([^_]*[^\s_])?' r'_(?!\w)'
view.basic_pattern_re = re.compile(basic_pattern)
#TODO: emoticons
# emoticons
emoticons_pattern = ''
if config.get('emoticons_theme'):
# When an emoticon is bordered by an alpha-numeric character it is NOT
# expanded. e.g., foo:) NO, foo :) YES, (brb) NO, (:)) YES, etc.
# We still allow multiple emoticons side-by-side like :P:P:P
# sort keys by length so :qwe emot is checked before :q
keys = interface.emoticons.keys()
keys.sort(interface.on_emoticon_sort)
emoticons_pattern_prematch = ''
emoticons_pattern_postmatch = ''
emoticon_length = 0
for emoticon in keys: # travel thru emoticons list
emoticon_escaped = re.escape(emoticon) # espace regexp metachars
emoticons_pattern += emoticon_escaped + '|'# | means or in regexp
if (emoticon_length != len(emoticon)):
# Build up expressions to match emoticons next to other emoticons
emoticons_pattern_prematch = emoticons_pattern_prematch[:-1] + ')|(?<='
emoticons_pattern_postmatch = emoticons_pattern_postmatch[:-1] + ')|(?='
emoticon_length = len(emoticon)
emoticons_pattern_prematch += emoticon_escaped + '|'
emoticons_pattern_postmatch += emoticon_escaped + '|'
# We match from our list of emoticons, but they must either have
# whitespace, or another emoticon next to it to match successfully
# [\w.] alphanumeric and dot (for not matching 8) in (2.8))
emoticons_pattern = '|' + \
try:
if config.get('emoticons_theme'):
# When an emoticon is bordered by an alpha-numeric character it is NOT
# expanded. e.g., foo:) NO, foo :) YES, (brb) NO, (:)) YES, etc.
# We still allow multiple emoticons side-by-side like :P:P:P
# sort keys by length so :qwe emot is checked before :q
keys = interface.emoticons.keys()
keys.sort(interface.on_emoticon_sort)
emoticons_pattern_prematch = ''
emoticons_pattern_postmatch = ''
emoticon_length = 0
for emoticon in keys: # travel thru emoticons list
emoticon_escaped = re.escape(emoticon) # espace regexp metachars
emoticons_pattern += emoticon_escaped + '|'# | means or in regexp
if (emoticon_length != len(emoticon)):
# Build up expressions to match emoticons next to other emoticons
emoticons_pattern_prematch = emoticons_pattern_prematch[:-1] + ')|(?<='
emoticons_pattern_postmatch = emoticons_pattern_postmatch[:-1] + ')|(?='
emoticon_length = len(emoticon)
emoticons_pattern_prematch += emoticon_escaped + '|'
emoticons_pattern_postmatch += emoticon_escaped + '|'
# We match from our list of emoticons, but they must either have
# whitespace, or another emoticon next to it to match successfully
# [\w.] alphanumeric and dot (for not matching 8) in (2.8))
emoticons_pattern = '|' + \
'(?:(?<![\w.]' + emoticons_pattern_prematch[:-1] + '))' + \
'(?:' + emoticons_pattern[:-1] + ')' + \
'(?:(?![\w.]' + emoticons_pattern_postmatch[:-1] + '))'
except:
pass
# because emoticons match later (in the string) they need to be after
# basic matches that may occur earlier
@ -230,10 +230,17 @@ def _parse_css_color(color):
return gtk.gdk.Color(r, g, b)
else:
return gtk.gdk.color_parse(color)
def style_iter(style):
return (map(lambda x:x.strip(),item.split(':', 1)) for item in style.split(';') if len(item.strip()))
class HtmlHandler(xml.sax.handler.ContentHandler):
"""A handler to display html to a gtk textview.
It keeps a stack of "style spans" (start/end element pairs)
and a stack of list counters, for nested lists.
"""
def __init__(self, textview, startiter):
xml.sax.handler.ContentHandler.__init__(self)
self.textbuf = textview.get_buffer()
@ -303,16 +310,20 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
frac, callback, args):
callback(allocation.width*frac, *args)
def _parse_length(self, value, font_relative, callback, *args):
def _parse_length(self, value, font_relative, block_relative, minl, maxl, callback, *args):
'''Parse/calc length, converting to pixels, calls callback(length, *args)
when the length is first computed or changes'''
if value.endswith('%'):
frac = float(value[:-1])/100
val = float(value[:-1])
sign = cmp(val,0)
# limits: 1% to 500%
val = sign*max(1,min(abs(val),500))
frac = val/100
if font_relative:
attrs = self._get_current_attributes()
font_size = attrs.font.get_size() / pango.SCALE
callback(frac*display_resolution*font_size, *args)
else:
elif block_relative:
# CSS says 'Percentage values: refer to width of the closest
# block-level ancestor'
# This is difficult/impossible to implement, so we use
@ -323,27 +334,42 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
self.textview.connect('size-allocate',
self.__parse_length_frac_size_allocate,
frac, callback, args)
else:
callback(frac, *args)
return
elif value.endswith('pt'): # points
callback(float(value[:-2])*display_resolution, *args)
val = float(value[:-2])
sign = cmp(val,0)
# validate length
val = sign*max(minl,min(abs(val*display_resolution),maxl))
if value.endswith('pt'): # points
callback(val*display_resolution, *args)
elif value.endswith('em'): # ems, the height of the element's font
elif value.endswith('em'): # ems, the width of the element's font
attrs = self._get_current_attributes()
font_size = attrs.font.get_size() / pango.SCALE
callback(float(value[:-2])*display_resolution*font_size, *args)
callback(val*display_resolution*font_size, *args)
elif value.endswith('ex'): # x-height, ~ the height of the letter 'x'
# FIXME: figure out how to calculate this correctly
# for now 'em' size is used as approximation
attrs = self._get_current_attributes()
font_size = attrs.font.get_size() / pango.SCALE
callback(float(value[:-2])*display_resolution*font_size, *args)
callback(val*display_resolution*font_size, *args)
elif value.endswith('px'): # pixels
callback(int(value[:-2]), *args)
callback(val, *args)
else:
warnings.warn('Unable to parse length value "%s"' % value)
try:
# TODO: isn't "no units" interpreted as pixels?
val = int(value)
sign = cmp(val,0)
# validate length
val = sign*max(minl,min(abs(val),maxl))
callback(val, *args)
except:
warnings.warn('Unable to parse length value "%s"' % value)
def __parse_font_size_cb(length, tag):
tag.set_property('size-points', length/display_resolution)
@ -352,7 +378,7 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
def _parse_style_display(self, tag, value):
if value == 'none':
tag.set_property('invisible','true')
#Fixme: display: block, inline
# FIXME: display: block, inline
def _parse_style_font_size(self, tag, value):
try:
@ -377,7 +403,8 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
if value == 'larger':
tag.set_property('scale', pango.SCALE_LARGE)
return
self._parse_length(value, True, self.__parse_font_size_cb, tag)
# font relative (5 ~ 4pt, 110 ~ 72pt)
self._parse_length(value, True, False, 5, 110, self.__parse_font_size_cb, tag)
def _parse_style_font_style(self, tag, value):
try:
@ -399,11 +426,13 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
#__frac_length_tag_cb = staticmethod(__frac_length_tag_cb)
def _parse_style_margin_left(self, tag, value):
self._parse_length(value, False, self.__frac_length_tag_cb,
# block relative
self._parse_length(value, False, True, 1, 1000, self.__frac_length_tag_cb,
tag, 'left-margin')
def _parse_style_margin_right(self, tag, value):
self._parse_length(value, False, self.__frac_length_tag_cb,
# block relative
self._parse_length(value, False, True, 1, 1000, self.__frac_length_tag_cb,
tag, 'right-margin')
def _parse_style_font_weight(self, tag, value):
@ -469,13 +498,32 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
tag.set_property('wrap_mode', gtk.WRAP_WORD)
elif value == 'nowrap':
tag.set_property('wrap_mode', gtk.WRAP_NONE)
def __length_tag_cb(self, value, tag, propname):
try:
tag.set_property(propname, value)
except:
gajim.log.warn( "Error with prop: " + propname + " for tag: " + str(tag))
def _parse_style_width(self, tag, value):
if value == 'auto':
return
self._parse_length(value, False, False, 1, 1000, self.__length_tag_cb,
tag, "width")
def _parse_style_height(self, tag, value):
if value == 'auto':
return
self._parse_length(value, False, False, 1, 1000, self.__length_tag_cb,
tag, "height")
# build a dictionary mapping styles to methods, for greater speed
__style_methods = dict()
for style in ['background-color', 'color', 'font-family', 'font-size',
'font-style', 'font-weight', 'margin-left', 'margin-right',
'text-align', 'text-decoration', 'white-space', 'display' ]:
'text-align', 'text-decoration', 'white-space', 'display',
'width', 'height' ]:
try:
method = locals()['_parse_style_%s' % style.replace('-', '_')]
except KeyError:
@ -489,6 +537,8 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
return [tag for tag in self.styles if tag is not None]
def _create_url(self, href, title, type_, id_):
'''Process a url tag.
'''
tag = self.textbuf.create_tag(id_)
if href and href[0] != '#':
tag.href = href
@ -501,6 +551,125 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
tag.title = title
return tag
def _process_img(self, attrs):
'''Process a img tag.
'''
try:
# Wait maximum 1s for connection
socket.setdefaulttimeout(1)
try:
f = urllib2.urlopen(attrs['src'])
except Exception, ex:
gajim.log.debug(str('Error loading image %s ' % attrs['src'] + ex))
pixbuf = None
alt = attrs.get('alt', 'Broken image')
else:
# Wait 0.1s between each byte
try:
f.fp._sock.fp._sock.settimeout(0.5)
except:
pass
# Max image size = 2 MB (to try to prevent DoS)
mem = ''
deadline = time.time() + 3
while True:
if time.time() > deadline:
gajim.log.debug(str('Timeout loading image %s ' % \
attrs['src'] + ex))
mem = ''
alt = attrs.get('alt', '')
if alt:
alt += '\n'
alt += _('Timeout loading image')
break
try:
temp = f.read(100)
except socket.timeout, ex:
gajim.log.debug('Timeout loading image %s ' % attrs['src'] + \
str(ex))
mem = ''
alt = attrs.get('alt', '')
if alt:
alt += '\n'
alt += _('Timeout loading image')
break
if temp:
mem += temp
else:
break
if len(mem) > 2*1024*1024:
alt = attrs.get('alt', '')
if alt:
alt += '\n'
alt += _('Image is too big')
break
pixbuf = None
if mem:
# Caveat: GdkPixbuf is known not to be safe to load
# images from network... this program is now potentially
# hackable ;)
loader = gtk.gdk.PixbufLoader()
dims = [0,0]
def height_cb(length):
dims[1] = length
def width_cb(length):
dims[0] = length
# process width and height attributes
w = attrs.get('width')
h = attrs.get('height')
# override with width and height styles
for attr, val in style_iter(attrs.get('style', '')):
if attr == 'width':
w = val
elif attr == 'height':
h = val
if w:
self._parse_length(w, False, False, 1, 1000, width_cb)
if h:
self._parse_length(h, False, False, 1, 1000, height_cb)
def set_size(pixbuf, w, h, dims):
'''FIXME: floats should be relative to the whole
textview, and resize with it. This needs new
pifbufs for every resize, gtk.gdk.Pixbuf.scale_simple
or similar.
'''
if type(dims[0]) == float:
dims[0] = int(dims[0]*w)
elif not dims[0]:
dims[0] = w
if type(dims[1]) == float:
dims[1] = int(dims[1]*h)
if not dims[1]:
dims[1] = h
loader.set_size(*dims)
if w or h:
loader.connect('size-prepared', set_size, dims)
loader.write(mem)
loader.close()
pixbuf = loader.get_pixbuf()
alt = attrs.get('alt', '')
if pixbuf is not None:
tags = self._get_style_tags()
if tags:
tmpmark = self.textbuf.create_mark(None, self.iter, True)
self.textbuf.insert_pixbuf(self.iter, pixbuf)
self.starting = False
if tags:
start = self.textbuf.get_iter_at_mark(tmpmark)
for tag in tags:
self.textbuf.apply_tag(tag, start, self.iter)
self.textbuf.delete_mark(tmpmark)
else:
self._insert_text('[IMG: %s]' % alt)
except Exception, ex:
gajim.log.error('Error loading image ' + str(ex))
pixbuf = None
alt = attrs.get('alt', 'Broken image')
try:
loader.close()
except:
pass
return pixbuf
def _begin_span(self, style, tag=None, id_=None):
if style is None:
@ -511,9 +680,9 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
tag = self.textbuf.create_tag(id_)
else:
tag = self.textbuf.create_tag() # we create anonymous tag
for attr, val in [item.split(':', 1) for item in style.split(';') if len(item.strip())]:
attr = attr.strip().lower()
val = val.strip()
for attr, val in style_iter(style):
attr = attr.lower()
val = val
try:
method = self.__style_methods[attr]
except KeyError:
@ -603,6 +772,7 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
self.text += content
return
if allwhitespace_rx.match(content) is not None and self._starts_line():
self.text += ' '
return
self.text += content
self.starting = False
@ -611,9 +781,8 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
def startElement(self, name, attrs):
self._flush_text()
klass = [i for i in attrs.get('class',' ').split(' ') if i]
style = attrs.get('style','')
style = ''
#Add styles defined for classes
#TODO: priority between class and style elements?
for k in klass:
if k in classes:
style += classes[k]
@ -641,9 +810,13 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
tag.is_anchor = True
elif name in LIST_ELEMS:
style += ';margin-left: 2em'
elif name == 'img':
tag = self._process_img(attrs)
if name in element_styles:
style += element_styles[name]
# so that explicit styles override implicit ones,
# we add the attribute last
style += ";"+attrs.get('style','')
if style == '':
style = None
self._begin_span(style, tag, id_)
@ -681,83 +854,7 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
elif name == 'dt':
if not self.starting:
self._jump_line()
elif name == 'img':
# Wait maximum 1s for connection
socket.setdefaulttimeout(1)
try:
f = urllib2.urlopen(attrs['src'])
except Exception, ex:
gajim.log.debug(str('Error loading image %s ' % attrs['src'] + ex))
pixbuf = None
alt = attrs.get('alt', 'Broken image')
else:
# Wait 10ms between each byte
try:
f.fp._sock.fp._sock.settimeout(0.1)
except:
pass
# Max image size = 2 MB (to try to prevent DoS) in Max 3s
mem = ''
deadline = time.time() + 3
while True:
if time.time() > deadline:
gajim.log.debug(str('Timeout loading image %s ' % \
attrs['src'] + ex))
mem = ''
alt = attrs.get('alt', '')
if alt:
alt += '\n'
alt += _('Timeout loading image')
break
try:
temp = f.read(100)
except socket.timeout, ex:
gajim.log.debug('Timeout loading image %s ' % attrs['src'] + \
str(ex))
mem = ''
alt = attrs.get('alt', '')
if alt:
alt += '\n'
alt += _('Timeout loading image')
break
if temp:
mem += temp
else:
break
if len(mem) > 2*1024*1024:
alt = attrs.get('alt', '')
if alt:
alt += '\n'
alt += _('Image is too big')
break
if mem:
# Caveat: GdkPixbuf is known not to be safe to load
# images from network... this program is now potentially
# hackable ;)
loader = gtk.gdk.PixbufLoader()
loader.write(mem)
loader.close()
pixbuf = loader.get_pixbuf()
else:
pixbuf = None
if pixbuf is not None:
tags = self._get_style_tags()
if tags:
tmpmark = self.textbuf.create_mark(None, self.iter, True)
self.textbuf.insert_pixbuf(self.iter, pixbuf)
if tags:
start = self.textbuf.get_iter_at_mark(tmpmark)
for tag in tags:
self.textbuf.apply_tag(tag, start, self.iter)
self.textbuf.delete_mark(tmpmark)
else:
self._insert_text('[IMG: %s]' % alt)
elif name == 'body' or name == 'html':
pass
elif name == 'a':
elif name in ('a', 'img', 'body', 'html'):
pass
elif name in INLINE:
pass
@ -807,10 +904,6 @@ class HtmlHandler(xml.sax.handler.ContentHandler):
# self.text = ' '
class HtmlTextView(gtk.TextView):
__gtype_name__ = 'HtmlTextView'
__gsignals__ = {
'url-clicked': (gobject.SIGNAL_RUN_LAST, None, (str, str)), # href, type
}
def __init__(self):
gobject.GObject.__init__(self)
@ -886,16 +979,39 @@ class HtmlTextView(gtk.TextView):
parser.setContentHandler(HtmlHandler(self, eob))
parser.parse(StringIO(html))
# too much space after :)
#if not eob.starts_line():
# buffer.insert(eob, '\n')
change_cursor = None
if __name__ == '__main__':
import os
from common import gajim
class log(object):
def debug(self, text):
print "debug:", text
def warn(self, text):
print "warn;", text
def error(self,text):
print "error;", text
gajim.log=log()
if gajim.config.get('emoticons_theme'):
print "emoticons"
htmlview = HtmlTextView()
path_to_file = os.path.join(gajim.DATA_DIR, 'pixmaps', 'muc_separator.png')
# use this for hr
htmlview.focus_out_line_pixbuf = gtk.gdk.pixbuf_new_from_file(path_to_file)
tooltip = tooltips.BaseTooltip()
def on_textview_motion_notify_event(widget, event):
'''change the cursor to a hand when we are over a mail or an url'''
@ -965,9 +1081,9 @@ if __name__ == '__main__':
<body xmlns='http://www.w3.org/1999/xhtml'>
<p style='text-align:center'>Hey, are you licensed to <a href='http://www.jabber.org/'>Jabber</a>?</p>
<p style='text-align:right'><img src='http://www.jabber.org/images/psa-license.jpg'
alt='A License to Jabber'
height='261'
width='537'/></p>
alt='A License to Jabber'
width='50%' height='50%'
/></p>
</body>
''')
htmlview.display_html('<hr />')
@ -992,8 +1108,9 @@ if __name__ == '__main__':
<li> One </li>
<li> Two is nested: <ul style='background-color:rgb(200,200,100)'>
<li> One </li>
<li> Two </li>
<li> Three </li>
<li style='font-size:50%'> Two </li>
<li style='font-size:200%'> Three </li>
<li style='font-size:9999pt'> Four </li>
</ul></li>
<li> Three </li></ol>
</body>