def print_depth_statistic(index):
html.div('<b>depth statistic</b>')
print """<table>
<tr><td>depth</td><td>pages</td></tr>"""
for s in index.get_depth_stat():
print '<tr><td>%i</td><td>%i</td></tr>' % s
print '</table>'
def print_common_statistic(index):
index_cnt, page_cnt, dome_cnt = index.len()
html.div('<b>common statistic</b>')
html.div('%i words in index', index_cnt)
html.div('%i pages indexed', page_cnt)
html.div('%i domains indexed', dome_cnt)
html.br()
html.div('<b>most common words</b>')
print '<table border=0>'
print '<tr><td>word</td><td>count</td></tr>'
for key, n in index.most_common(15):
print '<tr><td>%s</td><td>%i</td></tr>' % (key, n)
print '</table>'
html.header()
query = form.getvalue('q')
if not query:
query = ''
print """
<div/>
<a href="index.py"><b>search</b></a>
<a href="statistic.py">statistic</a>
<a href="web.py">web model</a>
<div><br/>
<form method="post">
<input name="q" value="%s">
<input name="s" type="submit" value="search">
</form>
""" % query
try:
index = IndexManager.get_proxy()
if query:
for url in index.lookup(query):
print '<div>'
html.a(url, url)
print '</div>'
except socket.error:
html.div("can't connect to DataStorage")
html.footer()
def print_domains(index):
html.div('<b>domains</b>')
for h in index.hosts():
print '<div>'
html.a('http://'+h, h)
print '</div>'