def need_update():
size = len(DOMAIN_CHARS)
squared = size * size
keys = (Prefix.all(keys_only=True).filter('length', 2)
.order('__key__').fetch(1000))
if len(keys) == 1000:
keys.extend(Prefix.all(keys_only=True).filter('length', 2)
.filter('__key__ >', keys[-1]).fetch(1000))
if len(keys) == squared:
oldest = (Prefix.all(keys_only=True).filter('length', 2)
.order('timestamp').fetch(100))
return [key.name() for key in oldest]
existing = set([key.name() for key in keys])
result = []
for c1 in DOMAIN_CHARS:
for c2 in DOMAIN_CHARS:
name = c1 + c2
if name not in existing:
result.append(name)
return result
def main():
for name in 'ab as ax ur we'.split():
prefix = Prefix(key_name=name, length=len(name))
prefix.count_domains()
print prefix
suffix = Suffix(key_name=name, length=len(name))
suffix.count_domains()
print suffix
for length in range(3, 7):
prefix = DotComPrefix(key_name='urcollection'[:length], length=length)
prefix.count_domains()
print prefix
suffix = DotComSuffix(key_name='urcollection'[-length:], length=length)
suffix.count_domains()
print suffix
def main():
one = {}
two = {}
total = 0
for c1 in DOMAIN_CHARS:
query = Prefix.all().filter('length', 2)
query.filter('__key__ >', db.Key.from_path('prefixes_prefix', c1))
next = chr(ord(c1) + 1)
query.filter('__key__ <', db.Key.from_path('prefixes_prefix', next))
one[c1] = 0
for prefix in query.fetch(100):
total += prefix.count
one[c1] += prefix.count
two[prefix.key().name()] = prefix.count
print 'TOTAL =', total
print
print_counts('ONE', one)
print
print_counts('TWO', two)
def main():
one = {}
two = {}
total = 0
for c1 in DOMAIN_CHARS:
query = Prefix.all().filter("length", 2)
query.filter("__key__ >", db.Key.from_path("prefixes_prefix", c1))
next = chr(ord(c1) + 1)
query.filter("__key__ <", db.Key.from_path("prefixes_prefix", next))
one[c1] = 0
for prefix in query.fetch(100):
total += prefix.count
one[c1] += prefix.count
two[prefix.key().name()] = prefix.count
print "TOTAL =", total
print
print_counts("ONE", one)
print
print_counts("TWO", two)
def main():
for name in 'ab as ax ur we'.split():
prefix = Prefix(key_name=name, length=len(name))
prefix.count_domains()
print prefix
suffix = Suffix(key_name=name, length=len(name))
suffix.count_domains()
print suffix
for length in range(3, 7):
prefix = DotComPrefix(key_name='urcollection'[:length], length=length)
prefix.count_domains()
print prefix
suffix = DotComSuffix(key_name='urcollection'[-length:], length=length)
suffix.count_domains()
print suffix
def index(request):
size = len(DOMAIN_CHARS)
squared = size * size
names = need_update()
random.shuffle(names)
names = names[:20]
initial = {'prefixes': ' '.join(names)}
prefix_form = PrefixForm(request.POST or None, initial=initial)
if prefix_form.is_valid():
prefixes = prefix_form.cleaned_data['prefixes'].split()
if not prefixes:
prefixes = names
for prefix in prefixes:
update_prefix(prefix)
return HttpResponseRedirect(request.path)
matrix = []
for letter in DOMAIN_CHARS:
matrix.append([None] * size)
domain_count = 0
prefix_count = 0
for prefix in Prefix.all().filter('length', 2):
x = DOMAIN_CHARS.index(prefix.key().name()[1])
y = DOMAIN_CHARS.index(prefix.key().name()[0])
matrix[y][x] = prefix.count
domain_count += prefix.count
prefix_count += 1
table_rows = []
sum_row = [0] * size
for y, letter in enumerate(DOMAIN_CHARS):
for x in range(size):
if matrix[y][x]:
sum_row[x] += matrix[y][x]
table_rows.append((letter, matrix[y],
sum([count for count in matrix[y] if count])))
if prefix_count:
domain_estimate = domain_count * squared / prefix_count
else:
domain_estimate = 'unknown'
letters = DOMAIN_CHARS
return render_to_response(request, 'prefixes/index.html', locals())
