def union(p1, p2):
if p1 is None and p2 is None:
return None
results = PostingsList()
while p1 is not None or p2 is not None:
if p1 is None:
add_until_end(p2, results)
p2 = None
elif p2 is None:
add_until_end(p1, results)
p1 = None
else:
if p1.entry < p2.entry:
results.add(p1.entry)
p1 = next_or_none(p1)
elif p1.entry == p2.entry:
results.add(p1.entry)
p1 = next_or_none(p1)
p2 = next_or_none(p2)
else:
results.add(p2.entry)
p2 = next_or_none(p2)
return results
def new_list(self):
if self._mode != 'w':
raise Exception("cannot create a new list without 'write' mode")
else:
new_postings_list = PostingsList()
self._postings.append(new_postings_list)
return (self._postings[-1], len(self._postings) - 1)
def list_at_offset(self, offset):
if self._mode == 'w':
if offset < len(self._postings):
l = self._postings[offset]
l.reset()
return l
else:
raise IndexError("offset out of bounds")
else:
self._f.seek(offset)
line = self._f.readline()[:-1] # strip out newline
p_list = PostingsList()
p_list.add_from_string(line)
if len(p_list._entries) == 0:
return None
return p_list
def __init__(self, file_name, mode='w'):
self._file_name = file_name
self._mode = mode
if self._mode == 'w':
self._postings = [PostingsList()]
else:
self._f = open(file_name)
def complement(p, c):
results = PostingsList()
while p is not None:
# c cannot overtake p as c is everything,
# and both move together when equal
if p.entry == c.entry:
p = next_or_none(p)
c = next_or_none(c)
else:
results.add(c.entry)
c = next_or_none(c)
if c is not None:
add_until_end(c, results)
return results
def intersect(p1, p2):
if p1 is None or p2 is None:
return None
results = PostingsList()
while p1 is not None and p2 is not None:
if p1.entry == p2.entry:
results.add(p1.entry)
p1 = next_or_none(p1)
p2 = next_or_none(p2)
elif p1.entry < p2.entry:
p1 = next_or_skip(p1, p2.entry)
else:
p2 = next_or_skip(p2, p1.entry)
return results