demo.py

from HTMLParser import HTMLParser

class MultiResultParser(HTMLParser):
    def __init__(self):
        HTMLParser.__init__(self)
        self.results = [[]]
        self.tag_type = None

    def handle_starttag(self, tag, attrs):
        attrs = dict(attrs)
        if tag == 'div' and attrs.get('class') == 'cell_input':
            self.tag_type = 'cell_input'
        elif tag == 'script' and attrs.get('type') == 'math/tex; mode=display':
            self.tag_type = 'latex'

    def handle_data(self, data):
        if self.tag_type == 'cell_input':
            self.results[-1].append(data)
        elif self.tag_type == 'latex':
            self.results[-1].append(data)
            self.results.append([])

        self.tag_type = None


case1 = """<ul>
<li><script type="math/tex; mode=display">4</script></li>
<li><script type="math/tex; mode=display" data-numeric="true" data-output-repr="-2 - 2*sqrt(3)*I" data-approximation="-2.0 - 3.46410161513775 i">-2 - 2 \sqrt{3} i</script></li>
<li><script type="math/tex; mode=display" data-numeric="true" data-output-repr="-2 + 2*sqrt(3)*I" data-approximation="-2.0 + 3.46410161513775 i">-2 + 2 \sqrt{3} i</script></li>
</ul>"""

case2 = """<ul>
<li>
<div class="cell_input">
integrate(exp(x)/(exp(2*x) + 1), x)
</div>
<script type="math/tex; mode=display">\operatorname{RootSum} {\left(4 z^{2} + 1, \Lambda {\left (i, i \log{\left (2 i + e^{x} \right )} \right )}\right)}</script>
</li>
<li>
<div class="cell_input">
sympy.integrals.manualintegrate(exp(x)/(exp(2*x) + 1), x)
</div>
<script type="math/tex; mode=display">\operatorname{atan}{\left (e^{x} \right )}</script>
</li>
</ul>"""

parser = MultiResultParser()
parser.feed(case1)
print "Case 1:"
print case1
print
print "Results:"
print parser.results

print
print

parser = MultiResultParser()
parser.feed(case2)
print "Case 2:"
print case2
print
print "Results:"
print parser.results

results.txt

Case 1:
<ul>
<li><script type="math/tex; mode=display">4</script></li>
<li><script type="math/tex; mode=display" data-numeric="true" data-output-repr="-2 - 2*sqrt(3)*I" data-approximation="-2.0 - 3.46410161513775 i">-2 - 2 \sqrt{3} i</script></li>
<li><script type="math/tex; mode=display" data-numeric="true" data-output-repr="-2 + 2*sqrt(3)*I" data-approximation="-2.0 + 3.46410161513775 i">-2 + 2 \sqrt{3} i</script></li>
</ul>

Results:
[['4'], ['-2 - 2 \\sqrt{3} i'], ['-2 + 2 \\sqrt{3} i'], []]


Case 2:
<ul>
<li>
<div class="cell_input">
integrate(exp(x)/(exp(2*x) + 1), x)
</div>
<script type="math/tex; mode=display">\operatorname{RootSum} {\left(4 z^{2} + 1, \Lambda {\left (i, i \log{\left (2 i ight)}</script>
</li>
<li>
<div class="cell_input">
sympy.integrals.manualintegrate(exp(x)/(exp(2*x) + 1), x)
</div>
ight )}</script>th/tex; mode=display">\operatorname{atan}{\left (e^{x} 
</li>
</ul>

Results:
[['\nintegrate(exp(x)/(exp(2*x) + 1), x)\n', '\\operatorname{RootSum} {\\left(4 z^{2} + 1, \\Lambda {\\left (i, i \\log{\\left (2 i + e^{x} \right )} \right )}\right)}'], ['\nsympy.integrals.manualintegrate(exp(x)/(exp(2*x) + 1), x)\n', '\\operatorname{atan}{\\left (e^{x} \right )}'], []]