parse.py

from lxml.html import parse

# this can be a full weburl as well
doc = parse('knives.html').getroot()

knives = {}

# start by iterating through every table cell
for td in doc.cssselect('td'):
    # If the cell has align=center, valign=top, and isn't blank
    if (td.get('align') == 'center' and td.get('valign') == 'top' and
        td.text_content().strip()):
        # This is the name of the knife
        name = td.cssselect('b')[0].text_content().strip()
        # descend further to find the price
        for new_td in td.cssselect('div table tbody tr td'):
            price = new_td.text_content().strip()
            if price.startswith('$'):
                # knives[name] = price[1:] <--- No dollar sign
                knives[name] = price

# Print the values we saved
for name, price in knives.iteritems():
    print "%s, %s" % (name, price)