gistfile1.py

import time
import os
import xapian
import uuid

uuid_blank = '00000000-0000-0000-0000-000000000000'
uuid_list = [str(uuid.uuid1()) for i in range(0,100000)]

unalloced = xapian.WritableDatabase('unalloced', xapian.DB_CREATE_OR_OPEN)
prealloced = xapian.WritableDatabase('prealloced', xapian.DB_CREATE_OR_OPEN)    

docs = []
c = 1000

for i in range(1,c):
    doc = xapian.Document()
    doc.add_boolean_term(str(i))
    unalloced.replace_document(i, doc)
    docs.append(doc)
    if i % 100 == 0:
        print 'unallocated', i
        unalloced.flush()

unalloced.close()

print 'flushing unalloc'

for i in range(1,c):
    doc = xapian.Document()
    doc.add_boolean_term(str(i))
    for j in range(1,10000):
        doc.add_value(j, uuid_blank)
    prealloced.replace_document(i, doc)
    if i % 100 == 0:
        print 'preallocated', i
        prealloced.flush()

prealloced.close()

print 'flushing prealloced'
time.sleep(1)

unalloced = xapian.WritableDatabase('unalloced', xapian.DB_CREATE_OR_OPEN)
prealloced = xapian.WritableDatabase('prealloced', xapian.DB_CREATE_OR_OPEN)    

for i in range(1,c):
    for j in range(1,10000):
        docs[i-1].add_value(j, uuid_list[j])

print time.time()
for i in range(1,c):
    unalloced.replace_document(i, docs[i-1])
unalloced.flush()
print time.time()
print "done with unalloced"

print time.time()
for i in range(1,c):
    prealloced.replace_document(i, docs[i-1])
prealloced.flush()
print time.time()
print "done with prealloced"